Deep-Discharge Process

Deep-discharge or Overdischarge
• Li-ion batteries connected in series are prone to be • SEM and XRD results for LiyNi1/3Co1/3Mn1/3O2 (NCM) cathode
overdischarged or deep-discharged. and graphite anode indicate that the overdischarge-induced
• Deep-discharge results in various side effects, such as internal short circuit (ISCr) is caused by Cu deposition on
capacity degradation and internal short circuit. electrodes, suggesting possible Cu collector dissolution at
• During overdischarge the Cu current collector of the cell is the voltage near to −12% SOC.
oxidized to Cu2+. • A significant voltage platform is observed at approximately
• Simultaneously, over-deintercalation of lithium at the anode −12% SOC, and ISCr is detected after the cell is
during overdischarge causes decomposition of SEI, and the overdischarged when passing the platform.
decomposition of SEI generates gases, including carbon
dioxide.
• New SEI films form on the anode when the cell is
recharged. The growth of SEI films can lead to degradation
of the electrochemical charge transfer processes in the
electrodes.
• The side reactions that occur during extreme overdischarge
result in the solid-state amorphization of the transition metal
compounds.
• The changes in the morphology of the components within
the Li-ion battery leads to capacity degradation.
• The dissolution of the Cu collector also affects the lifespan
of the battery.
• Deep-discharge results in various side effects, such as
capacity degradation and internal short circuit.
Voltage analysis during overdischarge.
• The voltage during overdischarge is shown in Fig.
• The overdischarge profile can be approximately divided into 3 stages according
to the characteristics of the voltage variations.
 In Stage I (−11%< SOC≤ 0%), the voltage dropped rapidly from 3.4V to −2.19V,
following a clear plateau at about 1V.
 In Stage II (−11%< SOC≤ −20%), the voltage underwent a monotonic gradual
increase asymptotically to 0V without fluctuations.
 The voltage drop in Stage I is caused by overdischarge leads to deintercalation
of Li+ from the anode, and insertion into the cathode.
 In Stage II, when the anode potential reaches approximately 3.4~3.5V anodic
corrosion of the Cu collector is triggered; the anode potential thus enters an
electrochemical reaction platform for the Cu dissolution.
 Cu+ dissolved in the electrolyte can travel through the separator and deposit on
the cathode; the cathode potential thus increases due to the reduction of Cu +.
 The overpotential for Cu dissolution can account for the voltage valley at
approximately −11% SOC.
 In Stage III, the electrochemical reactions of Cu dissolution and deposition
continue, and the internal short becomes more severe, with a decrease of
resistance of ISCr (RISCr). Thus, the voltage increased slowly asymptotically to
0V.
 Schematic Analysis of Over-discharge Process:
• Figure illustrates the process of Cu
dissolution during over-discharge and the
formation of the internal short circuit (ISCr)
induced by over-discharge.
• The internal short caused by Cu deposition
occurs after the cell is over-discharged to
SOC < -12% and becomes more severe
during the over-discharge process.

Figure: Copper dissolution and deposition during overdischarge and the formation of
internal short circuit.

Figure: The relationship between RISCr and SOC.

• The results suggest that ISCr occurs after the inflection point B at approximately −12% SOC, where the first voltage plateau is
located. The RISCr declines with a lower over-discharge SOC.
 SEM images of NMC/Graphite Tested Cells
• After testing, the cross-sectional SEM (a) (b)
images of the tested cell, one of after
over-discharge testing, and another of
after aging at 45 °C was observed.
• Evidence of the dissolution of the copper
current collector is apparent for the over-
discharged graphite cell.
• Energy‐dispersive X‐ray spectroscopy
Figure: (a) SEM image of a graphite/NMC cell after aging at 45 °C, (b) a graphite
(EDS) was also performed on each cell after over-discharge testing.
electrode as shown in Figure.
• These spectra show that the Copper was
present in the EDS spectra of the graphite
negative electrode after over-discharge.

Figure: EDS spectra of the cross‐sectioned electrode.

 Critical outcome of commercial lithium-ion cells under over-discharge conditions
• Over-discharge below the manufacturer specified low • A study performed in a LiCoO2-based pouch cell that was
voltage limit will result in the dissolution of the copper anodic overdischarged to 0.0 V, detect and quantify the gases
current collector. formed during overdischarge are Carbon dioxide, carbon
• This can deposit on internal cell components and induce an monoxide, and methane.
internal short. • It is proposed that the carbon dioxide and carbon monoxide
• The dissolution of copper from the anode current collector were produced from the decomposition of the SEI.
causes additional issues such as delamination of the anode • It has found that the cell which is discharge to 112% DOD,
and the deposition of copper on the surfaces of the cathode, could achieved full capacity on recharge, while cells that
separator and anode, which hinders the flow and were terminated beyond 114.5% DOD could not be
intercalation/deintercalation of lithium-ions during charge recharged back to full.
and discharge. • The authors of this study also note that increasing the
• This causes lithium ions to deposit as lithium metal rather capacity ratio of the anode to the cathode can delay the
than intercalating into the anode electrode, as the surface of DOD at which copper dissolution begins.
the anode is covered with copper. • Overdischarge of 18650 cells in a parallel arrangement was
• It has been demonstrated that the SEI layer could break studied by Nemanick et al.
down if the anode potential exceeds 3.5 V vs. Li/Li+. • Cells were overdischarged via successive cycling in which
• However, the anode potential can reach as high as 3.8 V 60% of the nominal capacity was discharged and only 90%
when over-discharging a Li-ion cell to 0.0 V. of the removed charge was returned during recharge in
• Thus, it is possible that both copper dissolution and each cycle.
breakdown of the SEI layer are causes of capacity loss • All cells appeared to develop shorts after 9 cycles.
during the overdischarge process. • Cross-sectional SEM confirmed that dendrites had formed
on the cathode surface and penetrated through the
separator, often visibly connecting through to the anode.
 Commercial Li-ion cells under over-discharge conditions
• Commercial Panasonic NCR18650B cylindrical cells.
• Rated capacity of these cells is 3350 mAh at 25 °C @ 0.2 C-rate.
• Cathode and anode active materials are LiNi 0.8Co0.15Al0.05O2 & graphite with Al
and Cu as current collectors, respectively.
• Separator is composed of PP-PE-PP with a layer of alumina (Al 2O3).
• Temperature measurements were recorded from a K-type thermocouple (Ω)
attached to the exterior of cell surface.
• The results shows that extreme over-discharge can lead to the dissolution of
copper from the anodic current collector, which can lead to severe capacity
loss.
• Metallic copper deposited on the surface of the cathode and anode, and the
cathode-facing separator surfaces.
• Cells were found to be prone to failure due to blockage of the cathode’s
reactive surface by the deposited copper.
• Formation of copper bridges leads to internally short the cell. • This indicates that internal shorting due to
• It is reported that heat generation occur during attempted cycling after over- the formation of copper dendrites in over-
discharge, as the blockage of electrode surfaces by copper prevents Li-ions discharged cells is not a catastrophic event
from intercalating and causes them to deposit as Li metal dendrites. until recharge is attempted.
• These dendrites can then cause localized heating, venting, and thermal • Internally shorted cells in a large bank of
runaway upon repeated cycling. cells can be dangerous because large
• In this also, a deep over-discharge is studied and no dangerously high amounts of heat are generated during the
temperatures were observed, even under extreme over-discharge conditions. attempted recharge of a shorted cell.
 Case Study of a commercial cell 4) Measured a.c. impedance of fully charged cells using
impedance analyzer between 10 kHz and 50 MHz.
• Effects of over-discharge on performance and thermal
stability of a commercially pouch cell with rated capacity 7.8 5) Measured thicknesses of fully charged cells using a
Ah were investigated. micrometer.
• This cell consisted of a graphite anode, polyethylene (PE) 6) Repeated steps 2–6 five times (i.e. cells were kept 15 days
separator, and a LiCoO2 cathode. at each of the prescribed low voltages and cycled 25
times). The five repetitions of EXPS-1 are identified as
• Its electrolyte was a mixture of LiPF 6 with organic solvents
“EXPS-2”.
ethylene-carbonate (EC) and diethyl-carbonate (DEC)
7) Finally, all cells were cycled 100 times between 4.2 and 3.0
Experimental protocol followed V @ 0.8 A (normal operation step) and then their a.c.
impedance and thickness were measured again using the
procedure in steps 4 and 5.
1) Discharged and charged cells five times between 4.2 and
3.0 V at 0.8 and 0.4 A, respectively. This is identified as
“cell conditioning step”. Findings
2) After cell conditioning, the cells were discharged to 2.3 V @  Results indicate that overdischarging between 2.0 and 0.5 V
0.8 A; rest 10 min and discharged further to either of 2.0, can lead to permanent capacity loss and adversely affects
1.5, 1.0, 0.5 or 0.0 V @ 10 mA. cycle-life of this Li-ion cell under normal operating
3) All cells were kept at their pre-fixed discharged voltage for 3 conditions.
days; then charged to 2.75 V @ 10 mA and further to 4.2 V  Overdischarge between 2.0- 1.5 V, and 1.0- 0.5 V led to
@ 0.4 A and finally discharged and charged five times using capacity losses of less than 2 and 18%, respectively.
the same current and voltage as in step 1. Steps 2 and 3  Same cells lost between 8 and 25% additional capacity
are identified as “EXPS-1”. when cycled 100 times between 4.2 and 3.0 V @ 0.8 A.
 Over-discharge between 2.0 and 0.0 V also has no impact on
thermal stability of this cell during overcharge conditions.
 Over-discharged between 2.0 and 1.0 V is close to or slightly
higher than a normal cell. However, both total heat generation
and thermo-chemical reactivity of cells over-discharged to 0.5
and 0.0 V are significantly lower than a normal cell, possibly
because of the degraded capacity of the cell.
 Behavior of this cell is unpredictable when over-discharged to
0.0 V. Of the seven cells tested, four cells managed to
withstand five consecutive repetitions of five cycles between
4.2 and 3.0 V and 3 days at 0.0 V, followed by 100 cycles of
charge/discharge between 4.2 and 3.0 V at 0.8 A.
 However, three cells failed operation at different stages of
repeated overdischarge to 0.0 V.
 Two of these cells exhibited no evidence of copper current
collector dissolution.
 The third cell, however, showed a considerable amount of
copper on the separator, with localized deposits greater on the
anode side than the cathode side.
 This is indicative of the fact that dissolved copper (Cu2+) can
indeed migrate through the separator from the anode side to
the cathode side and possibly cause an internal short
 Cell equalisation circuits
 Modern battery packs are made up of hundreds or  Over discharging: If the lower safety threshold a cell is
thousands of individual cells connected in different ignored the cell can become over discharged. In this
configurations of series and parallel in order to provide the mechanism cells exhibit thermal instability, permanent loss
necessary voltage (No. of series connections) and capacity of capacity and possible short circuit due to the copper
(No. of parallel connections). anode migrating across the cell to the cathode.
 Variations in the cells caused by the manufacturing process,  To avoid these problems, regular cell balancing is required.
charge discharge cycles and the temperature during The perfect solution would be described as an equalization
operation means that the series connected cells can circuit that has the following desirable characteristics:
become unbalanced in the amount of charge they are  Cheap.
storing. This leads to issues such as:  Scalable for large or differently configured battery packs.
 Undercharging: When a number of cells is left below its  Simple control method and implementation.
maximum capacity. Although not dangerous, but it limits
 Low switch stress.
the overall capacity of the pack resulting in reduced EV
driving range.  High efficiency with low thermal production.
 Overcharging: A very dangerous as cells forced to  High thermal resilience.
charge past their safety threshold results in large amounts  Able to equalize while charging or discharging i.e.,
of thermal energy produced which could result in chemical bidirectional.
explosion.
 High effectiveness, able to balance a large number of cells
 Under discharging: This is caused by a number of cells with minimal variation.
that have lower capacity than the others in the string.
 High speed, able to balance cells very quickly.
When these cells reach their lower safety threshold the
battery pack will believe it is fully discharged. The cells  Low circuit volume (doesn’t take up a lot of space).
that remain with charge cannot use this energy and  Not one circuit offers all of these attributes, although there
therefore overall capacity of the battery pack is lost. are many circuits that offer an interesting balance.
 Cell equalisation categorize in two groups
Passive equalization: Active equalisation:
• Passive equalization relies on resistance to dissipate excess • Active equalisation deals with the transfer of energy rather
energy from the high voltage cells within a series string. than the dissipation and is much more efficient.
• This form of equalisation is most popular in industry due to • In this case, energy from an overcharged cell or cells is
economical and simple control schemes. stored temporarily in a capacitor, an inductor or a
• Traditional passive equalisation either uses a fixed resistor transformer, and then discharged into an uncharged cells.
or a switched resistor. • In this, the circuit work on a shuttling, i.e., the energy is
• In fixed resistor, continuous bypass current limit the voltage. automatically transferred to and from adjacent cells and
allowing them to be balanced.
• This is not for LiB packs as cells can get overcharged.
• The equalisation speed of this method is dependent on the
• Switched resistor target the cells with the highest voltages. number of cells in the string and position.
• Once identified, a resistor is switched in and the cell • In cell to cell method voltage sensing is required and a
provides a current through the resistor. simple control system to order the cells from highest to
• Energy is dissipated until the cell is balanced with the rest of lowest SOC.
the cell in series. • Cells are directly targeted to transfer energy from one to
• Problems include very long equalisation time, excessive another. This method is quite slow, especially for large
heat produced. strings of series connected cells.
• Equalisation current must be kept low because of power • Cell to pack methods is when an overcharged cell
losses and TM, which leads to slow equalisation times. distributes its excess energy to the whole pack (or module).
• In a recent study, switched resistor circuit is replaced by the • This iterative process continues until the pack is balanced,
MOSFET driver circuit to control equalisation currents. This requiring sensing and a controller to select the overcharged
circuit is essentially turning the MOSFET into a variable cells.
resistor.
• Pack to cell method takes a small amount of energy from Double tiered switched capacitor
every cell and then uses that energy to charge the lowest
• By adding another tier of capacitors, cells that are not
energy cell in the string. Just like this is an iterative process.
directly connected together via a capacitor can now equalize
• The last category is multi cell to multi cell equalisation. through the second tier.
• This method again requires sensing and a complex • This has the effect of decreasing equalisation time.
controller, which can select one cell or multiple cells to
• Equalisation occurs through the same mechanism of
discharge, storing the energy temporarily.
charging and discharging capacitors by regular switching.
Capacitor based Cell Equalization Circuit • The difference with a single tiered system is, that in the
worst case scenario of the upper most cell and the lower
Switched capacitor most cell being unbalanced there would be half as much
• The capacitors are constantly switched between adjacent switching events needed for balancing to occur.
cells. • The main disadvantage of this topology is the increased
• If there is a difference in SOC between the two cells, the cost of components, size of the circuit in terms of volume
capacitor is charged by one cell and then discharged and compared to other techniques the balancing speed is
through the adjacent cell. still quite slow.

• As all cells are connected this eventually balances the

whole pack.
• Advantage is that you don’t need complicated control
systems nor sensing. Balancing can be achieved while
charging or discharging.
• Disadvantage is the slow balancing speed, the worst case
scenario being that the first cell in the string is unbalanced
compared to the last cell string. Fig.: Switched capacitor CEC.
Single switched capacitor Inductor based Cell Equalization Circuit
• The single switched capacitor works very differently from the
previous two capacitor methods.
Buck-boost
• The basic circuit as shown in Fig. uses inductors rather than
• In this configuration, cells are targeted by a simple control
capacitors as temporary energy storage.
algorithm rather than excess energy being automatically
shuttled from cell to cell. • This is cell to cell equalisation scheme where a cell with a
• high SOC will be switched in series with an inductor.
A high SOC cell is selected by the controller and capacitor is
switched into parallel so the capacitor charged up. • This charges the inductor until steady current is achieved.
• A low SOC cell is then selected and switched into parallel • The inductor will then discharge to an adjacent cell with a
with the charged capacitor and energy is transferred from lower SOC.
the capacitor to the low SOC cell. • There is no control method required as the circuit operates
• An example of a simple single switched capacitor circuit can continuously in two modes: 1st even switches turned on and
be seen in Fig. odd switches are off, 2nd even switches are turned off and
• all odd switches are turned on.
This configuration has the advantage that any cell in the
string can be used to balance any other cell in the string • Balancing can occur during both charging and discharging
rather than just adjacent or adjacent plus one cells. cycles, with no complicated controller needed.
• The circuit uses one capacitor and the cost of scaling up is • This is a cell to cell method, therefore equalisation speed is
very reasonable. an issue.
• Disadvantage is the very slow overall balancing speed. If • Traditionally inductors are more expensive and bulky than
only two cells are out of balance then it is very quick. capacitors.
• Disadvantage is when using large strings of series
connected cells the equalisation time is large similar to
single switched capacitor.
Flyback
• The flyback converter is an isolated version of the single
switched inductor.
• Instead of having a single inductor for energy storage two
inductors are coupled providing a primary side for energy
storage and a secondary side to provide isolation.
• Topologies are similar to a single transformer and a multi-
winding transformer.
• The flyback structure can provide cell to pack, pack to cell
and cell to pack to cell depending on bidirectional switching Fig.: Basic switched inductor “Cell equalization circuit”
and control.
• Efficiency is high and the current & voltage stresses are low.
• The main drawbacks, as with the transformer based circuits
are the large volume, cost and control complexity.

Quasi-Resonant
• It is quite similar with the buck boost.
• It is a cell to cell equaliser with high efficiency. With the
addition of a resonant LC tank to the circuit.
• By adding a resonance element to the circuit the component
number, and therefore, overall cost and the volume of the
circuit increases.
Fig.: Quasi-resonant “Cell equalization circuit”
Transformer Based Cell Equalization Circuit
• These are often very efficient with quick equalisation
speeds.
• Simple topologies are either cell to pack or pack to cell.
• However, with the use of bidirectional switching and a more
complex controller, multicell to multicell can be achieved.
• Transformer based circuits in general are very expensive
and highly complex to control as well as having a large Fig.: Multiple Transformer Cell Equalisation Circuit
circuit footprint.
• Disadvantage is each cell requires its own transformer, the
Multiple transformer
volume and cost of the solution is very high.
• In this topology shown in Figure each cells has its own • An improvement of this circuit is the ramp converter which
dedicated transformer. allows one transformer for every two cells.
• This method is said to be cell to pack, when a cell has • The operation is very similar to the multiple transformer,
excess charge switching allows the cells to provide a however, when the energy is going into the pack, half the
discharge current into its transformer where the energy is cycle is dedicated to charging even numbered cells with the
stored within the magnetic field. other half cycle dedicated to the odd numbered cells. This
• When the switch is turned off the stored energy is then configuration reduces circuit cost and volume.
redistributed into all other connected transformers, and
therefore, cells.
• This iterative process continues until all cells are at the
same SOC.
• Equalisation is quick and the current and voltage stresses
on the switches are very low.
Switched single transformer
• The transformer ratio is based on the number of cells
• In order to reduce the costs a single transformer can be
connected e.g., for eight cells each secondary winding has a
utilized with a switching mechanism.
ratio of 1/8. Therefore, scalability is an issue when
• This can be set up as cell to pack, pack to cell or with considering 96 cells within a series string.
bidirectional switches cell to pack to cell.
• This circuit again benefits from a modular design to reduce
• A large number of switches are required for this circuit, with the cost of the transformer.
an expensive and large transformer.
• This configuration is best utilized in small modules of 8–10
cells within a large battery pack.
• A coupling capacitor allows each adjacent module to then
become balanced.
Multi-secondary windings transformer
• This transformer topology is a hybrid of two previously
discussed and the most popular in research.
Fig.: Multi-winding transformer Cell Equalisation Circuit
• In this cells have a dedicated secondary winding, however,
all share the same primary as shown in Fig.
• The transformers themselves, are a subject of much
research to improve size, cost and performance.
• This configuration reduces the amount of windings
compared to the multi transformer circuit. This reduces the
costs dramatically.
Discussion
• A comparison is made of the basic circuits based on a • Following that trend is the volume of the circuits as well,
series of cells. where transformer and inductors take up a lot of space
• A numbers of switches, resistors, capacitors, inductors, compared to capacitors.
diodes, transformers and multiwinding transforms are used. • The thermal characteristics and reliability of inductors are
• Comparison are made based upon speed, cost, circuit generally better than capacitors, therefore for circuits
volume and control complexity. needing to operate under high temperatures inductors may
be the better choice.
• A chain structure can improve on this to the detriment of
circuit cost and volume.
• The converter based adjacent cell to cell schemes offer
better equalisation speeds than the basic switched
capacitor.
• The direct cell to cell equalisation times are also slow, in the
worst-case scenario all cells would be out of balance with
each other.
• Due to the time-shared nature of these schemes only one
cell can be balanced at a time leading to very slow
equalisation.
• The main advantages of these schemes are that the size
and cost of the circuit are usually kept low.
• Any circuit that requires transformers or a high number of
inductors are quite costly, whereas the circuits that are
based on capacitors are generally cheaper.
Cell Failure due to deep discharge