Technical CommentsChemistry

Response to Comment on “Cycling Li-O2 batteries via LiOH formation and decomposition”

See allHide authors and affiliations

Science  06 May 2016:
Vol. 352, Issue 6286, pp. 667
DOI: 10.1126/science.aaf1652


We described a lithium-oxygen (Li-O2) battery comprising a graphene electrode, a dimethoxyethane-based electrolyte, and H2O and lithium iodide (LiI) additives, lithium hydroxide (LiOH) being the predominant discharge product. We demonstrate, in contrast to the work of Shen et al., that the chemical reactivity between LiOH and the triiodide ion (I3) to form IO3 indicates that LiOH can be removed on charging; the electrodes do not clog, even after multiple cycles, confirming that solid products are reversibly removed.

We reported on a lithium-oxygen battery that formed LiOH as the predominant discharge product, formed from a reduced graphene oxide (rGO) electrode, a dimethoxyethane (DME)–based electrolyte, and H2O and LiI additives (1). On the basis of observed oxygen release, we suggested the following charge reaction.

4LiOH + 2I3 ↔ 4Li+ + O2 + 2H2O + 6I (1)

We also clearly stated, “We stress, however, that the equilibria that occur in the presence of oxygen, water, and iodine are complex, often involving a series of polyanions (including IO and its protonated form); further mechanistic studies are required to understand the role of these complex equilibria in the redox processes” (1). Some of these equilibria are as follows (2)3I ↔ I3 + 2e (2)2OH + I3 ↔ IO + H2O + 2I (3)or 2LiOH + I3 ↔ LiIO + LiI +I + H2O (4)The IO ion will then disproportionate, forming IO3

3LiIO ↔ LiIO3 + 2LiI (5)

Shen et al. (3) in their Comment make three major points, which we discuss below.

1) Shen et al. (3) argue that the I3-LiOH reaction to produce O2 on charge is not feasible, because the OH/O2 couple under standard conditions [3.84/3.42 V vs. Li+/Li in acidic/basic standard conditions, respectively (3)] is higher than the I/I3 couple (reaction 2, ~3.0 V in DME). First, removal of LiOH in lithium-oxygen batteries below 3.2 V has previously been observed—e.g., with ruthenium-based catalysts and tetraglyme (4)/dimethyl sulfoxide (5)–based electrolytes and added water. Because the catalyst does not alter the equilibrium voltage, the equilibrium itself must be altered. In another Technical Response (6), we argued that the free-energy change of reaction 1 on moving from an aqueous to nonaqueous electrolyte system cannot be ignored and will change the equilibrium potential. Of note, we have observed that LiOH can be quantitatively removed via reactions 2 to 5, I3 reacting to form IO3 and I and suggesting an alternative mechanism [see figure 1 in (6) for details], the rate depending strongly on the amount of water and LiTFSI concentrations present in the system (Fig. 1). Reaction 1 is a four-electron process and will be strongly affected by the nature of the electrode; further studies are required to determine the relative rates of reaction 1 versus reactions 3 to 5 and the potential role that catalysts may play in determining this.

Fig. 1

UV-visible spectra after the reaction of saturated LiOH in a 3 wt % water/DME solution. (A) Without added 0.1 M LiTFSI. (B) With added 0.1 M LiTFSI. Peak intensity changes confirm the chemical reaction between LiOH and I3 via reactions 2 to 5, to form IO3 and I (7).

2) Shen et al. (3) observed in a transparent cell that the electrolyte color darkens on cycling due to I3 accumulation. They concluded that the first discharge involves the formation of LiOH via O2 reduction at ~2.6 V [figure 1B in (3)] but that the subsequent charge process involves the direct oxidation of I to I3, rather than the removal of LiOH, subsequent discharge processes, also observed at ~2.6 V, being a combination of further LiOH formation and I3 reduction. Because our ultraviolet (UV)–visible experiments show that the kinetics for reactions 2 to 5 are highly dependent on the water content [figure 1 in (6)], we suggest that the conditions used in Shen et al.′s experiment, 5 A/g for a charge/discharge capacity of 1 Ah/g—i.e., a 12-min charge rate—may have been too fast to allow reactions 2 to 5 to go to completion, and thus I3 and LiOH accumulate. By contrast, we used a charge time ranging from up to ~20 hours at 1 A/gc for thin electrodes [e.g., figure 1 and figures S11(d) and S20 in (1)] to hundreds of hours for thick electrodes [e.g., figure S10 in (1), the amount of water content [up to ~50,000 parts per million or 5 weight % (wt %)] in our batteries also being higher than in Shen et al.’s. We indeed observed that I3 reacts with LiOH, transforming back to colorless IO3 and I after 14 hours (Fig. 1B).

The proposed mechanism by Shen et al. does not explain the experiment reported in figure S20 of (1). When the battery was fully discharged in the first cycle and then charged to the same capacity as the previous discharge, the capacity and voltage profile of the discharges in the following 11 cycles were nearly identical to that of the first cycle. We have consistently shown that the discharged product in the first cycle is quantitatively LiOH [figure 2 and figure S19 in (1)], the end of discharge being marked by either pore clogging or full coverage of the rGO surface by nonconductive LiOH. If during the first charge, LiOH is not effectively removed, it is difficult to propose a simple explanation for the subsequent multiple discharge processes, which show essentially the same discharge capacity, because no pore volume or bare rGO surface would be available for LiOH formation. Although we recognize that I3/I redox chemistry must occur in subsequent cycles, it cannot be the sole discharge reaction, because the voltage is always consistent with LiOH formation (or Li2O2 formation, this product forming at a similar voltage in the absence of water). Finally, we note that assuming that the battery operates solely via the I/I3 couple, the maximum theoretical capacity is 0.89 mAh for our LiI concentrations. Figures S10(b) and S11(d) in (1) present data showing a capacity of >7000 mAhg−1 on charge for a thick-electrode corresponding to 1.0 mAh and a thin-electrode cell that was cycled for data for 1000 cycles followed by 15 cycles at 22,000 mAh/g (0.22 mAh). Although redox shuttle mechanisms likely occur, LiOH removal must also be occurring.

3) Shen et al. argue that the small capacity obtained in figure S3 of our original paper “does not support what they claimed, because the LiI was originally in its reduced state, so there was nothing to be further reduced and the discharge capacity was thus very small” (1). Our experiments were, however, performed after a first initial charge. Thus, the discharge capacity reflects the total I3 concentration after this charge. Under the conditions used here, the capacity of the I3/I couple is consistently noticeably lower than that of the total I content in the cell, due to the slow diffusion of the ions to the rGO electrode, the capacity reflecting both I diffusivity and the charge rate [see page 8 of the supplementary materials for (1)].

We are aware of the multiple competing reactions in this battery system, the relative kinetics of the various reactions being further complicated because (i) the two key additives, water and iodine, react with the Li metal anode, and (ii) LiOH formation consumes water, the water concentration dropping during discharge. For example, as the water concentration drops, the battery may operate via Li2O2 formation; I3 can then oxidize Li2O2, releasing O2. Developing a scalable, practical device requires much further work to optimize this chemistry.


View Abstract

Navigate This Article