Elsevier

Electrochimica Acta

Volume 45, Issues 8–9, 3 January 2000, Pages 1193-1201
Electrochimica Acta

Electrochemical investigation of lithium aromatic sulfonyl imide salts

https://doi.org/10.1016/S0013-4686(99)00381-3Get rights and content

Abstract

The ionic behavior of an aromatic lithium sulfonyl imide, i.e. lithium bis(4nitrophenyl)sulfonylimide, has been investigated in an amorphous poly(oxyethylene) network and compared to that of usual lithium salts dissolved in the same host polymer. The NO2 groups which are substituents at both phenyl groups, are expected to induce a strong electron-withdrawing effect on the imide negative charge to provide large delocalization and therefore generate high charge carrier concentration. The conductivities observed are lower than those of lithium salts such as (CF3SO2)2NLi (LiTFSI) dissolved in the same host polymer. An important increase of Tg with salt concentration reflects a greater stiffness of the polymer induced by the aromatic rings, which reduces mobility, and might account for the lower conductivity. Interestingly, the aromatic imide exhibits higher cationic mobility with t+ ranging between 0.4 and 0.45, compared to 0.08–0.12 for the fluorinated imide. Cyclic voltammetry shows that lithium bis(4-nitrophenyl)sulfonylimide undergoes irreversible reduction before lithium plating, which has been attributed to the presence of nitro group.

Introduction

In the last decade we have investigated the electrochemical behavior of lithium trifluoromethyl sulfonyl imide, generally abbreviated as LiTFSI, in various host polymers such as polydioxolane [1], poly(oxyethylene) based networks [2], [3], [4], [5] and poly(oxypropylene) ones [6]. In addition to these experimental investigations ab initio calculations were performed on the imide anion and compared with triflate and a disubstituted carbanion (CF3SO2)2CH [7]. This study evidenced the extent of the negative charge delocalization in the three salts, in particular in the imide anion. More surprising was the absolute hardness η, calculated from the HOMO and LUMO values, using the Pearson [8] treatment based on Koopman’s theorem. Indeed, contrarily to our expectations, ab initio calculations revealed that the imide was the hardest base. Later we investigated the approach of the lithium cation and found that it was preferentially chelated by two oxygen atoms, each belonging to one SO2 group [9]. Obviously, a preferential interaction of the cation with the oxygen atoms rather than with the nitrogen one might explain the absolute hardness of the imide anion, which might therefore behave as triflate or perchlorate anions. Another feature of the imide anion is its great flexibility as shown by the conformational analysis which reveals a very low energy barrier [9]. Lastly, several authors reported a very low cationic transport in LiTFSI/polymer electrolytes [10], [11]. These determinations performed by an electrochemical method [12] were confirmed in the case of poly(oxypropylene) (PPO), by pulse field gradient NMR which showed a very fast imide anion diffusion with respect to the lithium cation [13], [14]. Thus anion flexibility and anionic mobility are probably among the most deciding factors which insure a high conductivity in a wide variety of dry polymer electrolytes based on LiTFSI. As for the ionization/dissociation of LiTFSI, comparative ab initio calculations shown that the percent of covalent bonding in the three previous lithium salts was in the same range and even slightly higher in LiTFSI. Of course these values are not experimental and might raise criticism as they do not allow to appreciate the solvent effect on the ion-pair dissociation. Eventually anion mobility and flexibility are decisive in LiTFSI based polymer electrolytes.

It appeared interesting to investigate in more detail another sulfonimide, namely lithium bis(4-nitrophenyl)sulfonylimide (LiNPSI), whose synthesis, characterization and conductivity behavior where previously reported by Reibel et al. [15]. This salt was mainly designed to provide large delocalization of the negative charge especially through the strong electron-withdrawing effect of the NO2 groups which are substituents on both phenyl groups. Additional features are a lower cost of synthesis and the absence of fluorine atoms, which might facilitate further recycling of lithium batteries prepared therefrom. This previous work [15] showed that LiNPSI dissolved in linear high molecular weight poly(oxyethylene) would neither decrease the crystallinity of POE nor slow down the crystallization rate, contrarily to LiTFSI, and therefore presents poor conductivities at room temperature. The electrochemical stability study was briefly initiated, but the transport numbers were not determined. Some preliminary conductivity results in a polyether–urethane network allowed to compare LiNPSI with some other lithium salts at close salt concentration and segmental mobility. However due to a high cross-link density, the differences in conductivity were levered off. Since then, more mobile networks have been prepared and much better results obtained [16].

To deepen the electrochemical investigation of LiNPSI, we have selected as host polymer a network obtained by free-radical polymerization of an unsaturated polycondensate, whose preparation has already been described [5]. The cross-linking allows very good mechanical properties for thin membranes exhibiting storage modulus ≥2 MPa on the rubbery plateau up to 100°C. In addition, it was shown that the double bond incorporation, leading to cross-linking, strongly decreases the crystallinity. Thus, such host polymers (i) with a regular POE spacing between the cross-link sites; (ii) amorphous at room temperature; (iii) able to endure high temperature without creeping, are useful materials for physico-chemical approaches on polymer/salt interactions. Lastly we already investigated a wide variety of salts, including single-ion conductors, in these quasi-model networks, allowing thus useful comparisons.

Section snippets

Host polymer

The linear unsaturated polycondensates were prepared by a Williamson type polycondensation [5]. This polycondensation was performed on α, ω-dihydroxyoligo(oxyethylene) Mn=1000 g mol−1. The polymer thus obtained was noted LPC1000. The resulting polymer is purified by ultra-fltration in order to remove any material with a molecular weight lower than 10 000 g mol−1.

Salt

The lithium bis(4-nitrophenyl)sulfonylimide was prepared as previously described [15]. The salt was dried at 80°C for 2 days, and then

Ionic conductivities

Fig. 1 presents Arrhenius plots for several NPC1000/LiNPSI electrolytes. The best conductivity at 30°C, 8×10−6 S cm−1, is reached for a rather low salt concentration O/Li=30. But above 70°C, the composition O/Li=22, gives more conductive material reaching 8×10−5 S cm−1 at 90°C. Fig. 2 compares these conductivities with those obtained for a linear high molecular weight POE (Mn=9×105) at O/Li=30 concentration, which was reported to give the maximum conductivity values [15]. Due to the amorphous

Conclusion

The design of lithium bis(4-nitrophenyl)sulfonylimide (LiNPSI) was to provide a large delocalization of the negative charge due to the strong electron-withdrawing effect of the aromatic nitro groups. Thus an effect quite similar to CF3 in LiTFSI could be expected. However, the conductivities of LiNPSI are notably lower than those of LiTFSI. This can be attributed to the stiffness of the anion induced by the aromatic groups, reducing the polymer segment mobility, as shown by the higher glass

References (18)

  • S. Sylla et al.

    Electrochim. Acta

    (1992)
  • D. Benrabah et al.

    Electrochim. Acta

    (1992)
  • F. Alloin et al.

    Electrochim. Acta

    (1998)
  • C. Roux et al.

    Solid State Ion.

    (1994)
  • D. Benrabah et al.

    Electrochim. Acta

    (1995)
  • M. Watanabe et al.

    Solid State Ion.

    (1995)
  • F. Alloin et al.

    J. Power Sour.

    (1997)
  • J. Evans et al.

    Polymer

    (1987)
  • C. Roux et al.

    Electrochim. Acta

    (1998)
There are more references available in the full text version of this article.

Cited by (0)

View full text