The rapid expansion of ocean observations made from long-range autonomous ocean vehicles was made possible by satellite communication and locating, low-power sensors and digital controllers, and batteries with high energy density. Today’s operations are sustained by rechargeable (secondary) lithium-ion batteries with energy densities (at room temperature) near 0.7 MJ kg−1 and primary lithium batteries providing up to 2 MJ kg−1. Significant increases of range and speed of underwater vehicles or the duration and depth of profiling floats will require power sources with higher energy density.
These batteries have substantial potential advantages for long-duration ocean vehicles. Their high energy density (up to 4 MJ kg−1) is particularly valuable for neutrally buoyant vehicles. High energy/mass supports high performance, as the buoyancy and volume needed to carry a fixed energy decreases. The intrinsic high energy density of seawater batteries is amplified by vehicle design because the batteries do not require pressure protection, thereby reducing the size and weight of the vehicle pressure case. Surprisingly, these lithium batteries are safe because in the PLE lithium is sandwiched between water-impermeable windows that prevent its rapid oxidation. Even if these windows are broken completely, only a thin sliver of lithium is exposed and the reaction rate is, and remains, moderate. In laboratory handling we damaged two PLEs, one by breaking a window and another by damaging the seal connecting the two windows. In each case, submersion in water led to a small bubble (presumable H2) with diameter < 1 mm, escaping every 1–2 s.
At the same time, lithium-seawater batteries present challenges to tapping their potential. These cannot be addressed until the batteries have been tested in the ocean. The purpose of this note is to report the results of laboratory and sea tests of PolyPlus seawater batteries that are optimized to reduce oxygen at the cathode. At the outset, there were three main areas of concern.
a. Mass transfer at the electrodes
Growth of marine organisms on surfaces is endemic to the ocean and over weekly time scales can lead to buildup of slimy coatings and small solid organisms (e.g., barnacles) adhering to most surfaces. This is a serious threat to seawater batteries that depend on good mass transfer to flush electrode surfaces. Biofouling protection should help but soluble poisons could interfere with the electrochemical reactions.
c. Mechanical damage from pressure cycling
Profiling floats and underwater gliders operate by cycling over a range of depths. This high-pressure cycling might cause mechanical damage to the solid-electrolyte window of the anode. These windows would be broken under pressure if irregularities in the neighboring lithium applied concentrated stresses to them. For this reason, the lithium is immersed in a liquid electrolyte that equalizes stress on the windows. Nevertheless, large lithium irregularities could still cause stress concentrations, or gasses dissolved in this liquid under high pressure could come out of solution when the pressure decreases, expanding and rupturing the PLE.
The tests reported here are intended to quantify the field performance of the PolyPlus cells in the face of mass transfer issues, biofouling, and all sources of damage, including pressure cycling.
2. The field trial setup
The sea trial tested a pair of PolyPlus PLE cells with cathodes optimized for oxygen reduction. Each cell had an anode with a total (two sided) exposed area of 970 mm2 sandwiched between a pair of cathodes composed of a 0.1-mm-thick feltlike-cloth material. The electrodes are mounted in light plastic frames and separated by gaps for seawater. Figure 1 shows two views of a cell with its underwater connector. Two cells, with gaps of 6.4 and 9.4 mm, were tested simultaneously in a test package carried by a glider. Stripped of cables and connectors, each cell’s mass was near 10 g and contained enough lithium to deliver 2.5 Ah or about 31 kJ (8.7 Wh) at a potential of 3.8 V. A primary test objective was to determine the power that could be delivered by oxygen reduction at ambient oxygen concentrations and achievable mass transfer rates.
In field trials, each cell with its underwater connector (Fig. 1) was mounted on a fiber-reinforced plastic (G10) fin with both cell faces fully exposed (Fig. 2a). The fin was attached to the stern of a Spray underwater glider (Sherman et al. 2001) with the fins and cells pointing downward, parallel to the flow (Fig. 2b). Putting the cells below the glider hull kept them generally submerged to minimize fouling by surface-active materials and kept the space between electrodes filled with seawater to minimize air voids that could focus current flow into hot spots, causing cathodic precipitation. Spray operates with some flow crossing the hull to produce lift. When the glider ascends, the cells mounted below the hull are in the vehicle’s wake and mass transfer to the electrodes is decreased.
The two cells were operated as an independent test package, electrically connected to Spray only by signal lines passing commands from the glider controller and relaying data back for recording and transmission to shore. A signal caused each cell to be connected to a resistive load of 1, 3, 6, or 40 kΩ. Voltage across the load and hence cell current were measured at 8-s intervals. While there was no direct connection between the two cells, current could flow between them through their common seawater electrolyte. To minimize the effects of current flow between the two cells, data are reported only at times when both cells had the same load. Near the surface and the bottom of each dive, the glider slows significantly and mass transfer declines. To minimize potentially irreversible precipitation on the cathodes, the cells were switched to the minimal-drain 40-kΩ load as the vehicle slowed. Some current density is needed to keep sodium ions from infiltrating the anode windows and degrading their performance.
Since cell performance depends on electrode–electrolyte mass transfer, laboratory tests were initially carried out to estimate this transfer by moving cells through a large body of quiescent seawater. These indicated that a cell held with its electrodes parallel to a flow of 0.15 m s−1 or greater had adequate mass transfer to supply 3 mA (0.3 mA cm−2 of electrode area) at a voltage greater than 2.9 V if the seawater is oxygen saturated around 20°C. In the field, Spray moved at a speed near 0.3 m s−1 with a glide ratio near 3:1.
Spray SN17 was deployed on 21 March 2012 offshore of San Diego, California, carrying the lithium-seawater battery package and Sea-Bird Electronics conductivity–temperature–pressure (SBE41CP) and oxygen (SBE43F) sensors to profile salinity (S), temperature (T), and oxygen (O2). The vehicle operated for 69 days within 30 nm of 33°N, 117°50′W, where the mean oxygen profile was roughly constant at 5.7 mL L−1 (~0.24 mM) above 15-m depth, fell linearly to 3 mL L−1 at 50 m, and then declined exponentially to 0.4 mL L−1 at 400 m. Thus, performance was tested over an O(10) range of oxygen concentrations on each profile. Scatter about the mean oxygen profile was about ±0.5 mL L−1 near the surface, reached a maximum near ±1 mL L−1 at 50 m, and declined gradually to around ±0.1 mL L−1 below 300 m.
Spring is the most productive season in the operational region and typically results in the greatest fouling of gliders. The glider dove from the surface to about 400 m every 2–3 h. It maintained fairly constant horizontal and vertical velocities of 0.25 and 0.1 m s−1, respectively, while descending and ascending. Through-water motion nearly stopped during the transition from diving to ascending and when on the surface for positioning and data relay. The field data from this effort are a mixture of environmental and engineering measurements and are available to readers (Sherman and Davis 2016, unpublished manuscript).
3. Field results
Importantly, during the entire test, when the two independent cells were operated under the same loads, their voltages tracked closely. Skipping the first minute after a load change, when voltages changed rapidly, the rms cell difference was less than 2.5 mV under all load combinations, speeds, and glide slopes (up, down, and level). This indicates that 1) the cells were not damaged by pressure cycling, 2) the difference in interelectrode spacing between the cells was unimportant to voltage, and 3) the performance of neither cell was seriously affected by biofouling. After 729 dives to 400 m and 250 dives to 250 m, no damage or performance degradation of the cells was observed. This is convincing evidence that the PLE is robust to pressure cycling.
The biggest design challenge to using seawater batteries is providing adequate cell flushing so that reactants reach, and products leave, the electrodes rapidly. This is necessary to minimize the concentration polarization induced by current flow. Our tests were designed to see how serious this problem is for the PolyPlus lithium-seawater cell. As described above, each cell was reasonably well exposed to the free-stream velocity, which was near 0.3 m s−1 during profiling. Profiling exposed the cells to a range of bulk oxygen concentrations. Various test schedules for varying the cell load, vehicle speed, and depth were tried. The most illuminating started with the cell flushed by holding it with a 40-kΩ load for 1−5 min before the load was switched to a lower resistance while the vehicle ascended or descended. Cell voltage was tracked to see how polarization developed at different bulk oxygen concentrations.
Figure 3 shows the cell voltage when the load was switched from 40 to 6 kΩ as the glider motion changed from slow ascent to descent; only a part of the complete ascent and descent records are shown. Under a 40-kΩ load, cell voltage varied with the oxygen concentration (color codes), ranging from 3.52 at depth to 3.6 V near the surface, but the [O2]-versus-voltage relation was imperfect; some high [O2] dives had quite low voltage, presumably because of poor flushing.
In the first 8 s after the load increased, voltage abruptly fell 0.15 V. Some of this is due to internal resistance (in the water path and anode window) and some is due to rapid concentration polarization. Most of the subsequent voltage decline is from polarization, occurs within 50 s, and puts cell voltage in the 3.29–3.44-V range. The current-driven polarization is greatest at low bulk [O2], where it is about 0.2 V. Here, too, high [O2] is associated with high voltage but imperfectly. After 50 s, polarization appears to have equilibrated and the subsequent slow decline reflects lower [O2] as the glider descends into oxygen-poor water.
Conditions in Fig. 4 are identical to those in Fig. 3, except that the sense of vertical glider motion is reversed; the high load is applied as the glider ascends. Because the data shown are from the beginning of a long ascent, the bulk [O2] levels are lower than in Fig. 3. The preload-transition voltages, the immediate 0.15-V drop following a load change, and the first 10–20 s under a higher load are very similar to Fig. 3. But when ascending in moderate or low [O2], the polarization continues to increase longer and becomes much greater than in Fig. 3. Ascending also puts the cells in the wake of hull cross-flow, apparently contributing to the dramatic increase of current polarization (to 0.9 V) and time constant (to 300 s) compared with Fig. 3.
Figure 5 shows a load-transition at low oxygen concentration. The cathode ceased reducing oxygen by half-reaction (1b) and began evolving hydrogen by (1c). The conditions in this figure are as in Fig. 4, except that the second load is 1 kΩ. Before transition, the cell produced 0.007 mA cm−2 at 2.8 V; after the load changed, the current increased to 0.23 mA cm−2 at 2.3 V. The power density of 0.5 mW cm−2 is over half the highest achieved with oxygen reduction, also with a 1-kΩ load but at higher [O2].
Equilibrated performance under various loads is presented in Fig. 6, which shows scatterplots of cell voltage versus [O2]B (Fig. 6a) and cell power density versus [O2]B (Fig. 6b). Cell voltage is a nonlinear function of [O2] and load, with high [O2] and light loads favoring high voltage. After accounting for the variation of ambient oxygen with depth, the load dependence of voltage is seen to be mainly a result of limited mass transfer. Cell power is simpler because polarization decreases voltage as current increases, reducing power variability at a given load unless the cell shifts to a hydrolysis reaction.
4. Biofouling and precipitation
The greatest utility of a low-power seawater battery would be in slow, long-duration vehicles, so it is important that the battery resist biofouling that might inhibit cell flushing. Our sea test was in an area of high biological productivity during its peak season, so it is not surprising that biofouling by a diverse collection of biota, many of which we cannot identify, was observed. In a similar way, cathode reactions produce hydroxide ions that encourage precipitation of low-solubility salts, like calcium and magnesium carbonates and hydroxides. This leaves deposits on, and in, the weave of the cell cathodes. PolyPlus has warned that these deposits may interfere with the cathode’s ability to catalyze oxygen reduction.
The cell mounts and frames collected several juvenile barnacles. Indeed, cell 75 had a 2-cm-long peduncle extending along an edge of one interelectrode gap, significantly impeding flow through that gap. Inside the cells, fouling biota and precipitates were common. Photographs of the two cells after their 69-day cruise are shown. These pictures give other nonbiologists an impression of how severe biofouling was and, in conjunction with the performance descriptions above, how insensitive cell performance is to biofouling and precipitates.
Figure 7 shows significant fouling of cell 76 in the 9.4-mm gap between the anode and the lower cathode. This figure has three significant biological connections across this interelectrode gap. The largest is a white “pillar” of nearly constant ~1-cm diameter, attached to the edges of the two electrodes. A thinner and darker biostructure connects the anode to the cathode and, apparently, to a small animal on the cathode. The third connection, just to the right of the pillar, is smaller, with the appearance of a stalactite attached to the anode and extending diagonally to connect with the pillar and the anode. While all these appear to short the battery, performance indicates the short has high impedance and performance is little affected. This may reflect the limited conductance of the anode windows.
Other depositions on cell 76 are shown in Fig. 8. The upper cathode is covered by a complex pattern of deposits that are widespread and, in spots, hundreds of nanometers thick. These may be examples of high-pH precipitates, biological deposition, or a mixture of both. A prominent stalactite is connected to a different type of deposit that covers a significant part of the anode. This is of concern because it could significantly reduce the anode area and likely raises cell impedance.
Figure 9 shows a very different “ropey” biofouling on cell 75 that is of a type often seen in our tests. Another example of this fouling is seen on the cathode frame in Fig. 2a. “Ropes” are seen on electrode surfaces (both anodes and cathodes) but seem to favor surrounding surfaces like frames. A concern is that, given more time, the ropes might become a significant barrier to flushing of the cell’s 6.4-mm interelectrode gaps and degrade cell performance.
Figure 10 shows the fine structure of one of cell 75’s cathodes with mineralized deposits embedded in its rough surface. The cathode, critical for cell voltage, is a weave of material selected to promote the oxygen-reduction reaction. The deposits are salt precipitates formed when the cathode reaction alkalized the surrounding water and are a concern for long-term cathode performance. Nevertheless, performance of our cells was not substantially degraded in over 2 months of operation, much of it at moderately high current densities over 0.2 mA cm−2.
The above-mentioned illustrations show that biofouling and precipitation had significant physical effects on our test cells. It seems likely that covering the electrode frames and other plastic components of the cells with a low-solubility biocide like copper could reduce biofouling. Even without protection, cells operating in a very productive area hosted a diverse collection of biota and precipitates without measurable change in cell performance over 2-plus months. As with all sensors operating in seawater, biofouling is a concern but the PolyPlus cells seem reasonably insensitive to some impressive attached biota.
Of the study’s three primary objectives, we have clear-cut results for two. The lithium-seawater batteries were pressure cycled nearly 2000 times, mostly to 400 m, without observable damage or degradation. The PolyPlus Protected Lithium Electrodes appear robust to pressure cycling even as the enclosed lithium is consumed.
In a 69-day mission in the springtime California Current, the cells were subject to significant biofouling as documented in the given photographs, perhaps more than the Spray hull itself. Nevertheless, they continued to function without measurable degradation. Fouling certainly remains a primary design consideration, but with appropriate antifouling techniques, one can hope for several months of operation without serious performance degradation.
Describing the effects of limited mass transfer on cell performance is more complex. The cells work best when the cathode reaction is oxygen reduction, dissolved oxygen concentration is above 1 mL L−1, and mass transfer is good (during descent for our bottom-mounted configuration). Power densities of O(9) W m−2 were achieved at voltages above 2.9 V. When mass transfer is restricted (during ascent), this high power was possible only for [O2] > 5 mL L−1. At moderate current densities of O(0.1) mA cm−2, batteries work well for [O2] down to 0.5 mL L−1 if cell flushing is good. Even when oxygen and cell flushing are inadequate to support the oxygen-reduction reaction, a power density of 5 W m−2 was obtained when the cathode reaction switched to hydrogen production. At the loads tested, this reaction seemed little influenced by mass transfer. We have, however, been warned that the cathode may cease to be effective in oxygen reduction after it is used for long periods of hydrogen evolution.
In our minds, three questions must be answered before lithium-seawater batteries will significantly improve the speed or duration of long-lived neutral buoyancy vehicles:
Will the economics of cell production evolve to make seawater batteries more attractive than using a larger vehicle powered with conventional primary lithium batteries?
Can a cathode deliver good performance over long periods while alternatively reducing oxygen and evolving hydrogen? If not, for which reaction should the lithium seawater cells be optimized?
Can mass transfer be maintained or improved without adding excessive drag when large batteries are built? It may, for example, be better to use a hydrogen-evolution cell with minimal flushing than a higher-power oxygen-reduction cell that doubles vehicle drag.
While seawater batteries can increase glider duration and speed, drag must be kept low to do it. Just as engine cooling is a significant source of drag for aircraft powered by internal-combustion engines, flushing the cells of lithium seawater batteries is a major source of drag for gliders. Each PolyPlus cell, with an ~7-cm2 end-on area, delivers about 5 mW. To provide Spray’s O(1) W average power, 200 cells would be required; their total end-on area would be several times that of Spray. It seems that only a “fresh start” glider design will be able to carry the necessary cells at an acceptable drag cost. One strategy is to build the cells into well-exposed parts of the glider (lifting surfaces and hull)—limiting skin drag will be a challenge. Or a pump forcing convection through an array of cells inside a streamlined structure might be used, but an adequate pump will use significant energy. Although challenging, the performance benefits of solving this flushing–drag problem are great.
This paper is based on work supported by the Office of Naval Research under Award N00014-13-1-0455. Any opinions, findings, conclusions, or recommendations expressed in it are those of the authors and do not necessarily reflect the views of the Office of Naval Research. We want to particularly thank our program manager Terri Paluszkiewicz for her vision in supporting this high-risk effort; Steve Visco and Eugene Nimon of PolyPlus Battery Company for their generosity and patience in helping us; and Jim Bellingham of Woods Hole Oceanographic Institution for involving us in this fascinating opportunity.
Sherman, J. T., R. E. Davis, W. B. Owens, and J. Valdes, 2001: The autonomous underwater glider “Spray.” IEEE J. Oceanic Eng., 26, 437–446, doi:10.1109/48.972076.
- Search Google Scholar
- Export Citation