Doped Lithium transition metal Oxides containing Sulfur

The invention relates to a powderous lithium transition metal oxide, used as active cathode
material in rechargeable lithium batteries. More particularly, in Li(Mn-Ni-Co)O2 type
compounds containing sulfur the addition of certain amounts of elements like Ca, La, Y,
Sr, Ce or Zr optimizes the electrochemical and safety characteristics of the cathode
material.
LiCoO2 is the most widely applied cathode material for rechargeable batteries. However,
there exists a strong pressure to replace it by other materials for particular reasons.
Currently, scarce resources of cobalt and fear of high prices accelerate this trend. Besides
LiFePO4 and Li-Mn-spinel, which both suffer from much lower energy density, LfNiO2
based layered cathode materials and Li(Mn-Ni-Co)O2 based layered cathode materials are
the most likely candidates to replace LiCoO2 in commercial battery applications. Today it
is basically known that any composition Li[LixMi.x]O2 with M=Mn, Ni, Co within the
quarternary system Li[Li1/3Mn2/3]O2 - LiCoO2 - LiNiO2 - LiNi0.5Mn0.5O2 exists as a
layered phase, and in most cases is electrochemically active.
Even this quarternary system is to be seen as a simplified model because it does not take
into account further phenomena like the possibility of cation mixing. One type of cation
mixing is known from LiNiO2 where some nickel is misplaced on lithium sites of the r-3m
layered crystal structure, a more realistic formula is approximated as {Li1-xNix}[Ni]O2. It
is also known that Li1+xM'-xO2 with M=Mn1/3Ni1/3Co1/3 is better written as
{Li1+yMy}[LizM'-2]O2.
As a result, layered Li(Mn-Ni-Co)O2 phases which are of interest for battery cathode
materials belong to the quarternary (according Gibbs phases rule) subspace of the 5
dimensional thermodynamic system LiNiO2 - {Li1-aNia}NiO2 - Li[Li1/3Mn2/3]O2- LiCoO2
- LiNiO2- Most of the phases within this triangle are electrochemically active.
By very basic thermodynamic reasons, if further parameters are included (like oxygen
particle pressure or temperature) the numbers of dimensions might increase further to 5 or
6, to explain phenonema like dependence of cation mixing of a given composition as

function of temperature, or the existence of vacancies (oxygen or cationic) as function of
temperature and oxygen pressure as observed for LiCoO2.
This has not even taken account for further dopants, which might fit into the crystal
structure, like Mg, Al, Cr, Ti; such doping introducing further degrees of freedom, adding
more dimensions to the already complex thermodynamic system.
Since many years it is known that the layered structure of LiNiO2 can be stabilized, and
electrochemical properties can be improved if Ni is replaced by Mn or Co, resulting in
LiNi1-xMnxO2 and LiNi1-xCoO2. Quite soon it was discovered that Mn and Co can be co-
doped, resulting in layered Li(Ni-Mn-Co)O2 phases LiNi1-x-yMnxCoyO2. So JP3244314
(Sanyo) claims LiaMbNicCOdOe covering a wide range of metal compositions.
It was also discovered quite early that Al can replace Ni. So, already in the early and
middle nineties there exist many patent with claims like LixNi1-a-bMlaM2bO2 where
generally x is near to unity, M' is transition metal and M2 a further dopant like aluminum.
Examples typically focus on LiNiO2 based materials (say a+b < 0.4), and can be found in
JP3897387, JP3362583, JP 3653409 or JP3561607, the latter disclosing
LiaCobMncMdN1(b+c+d)O2 with (K a<1.2,0.1<=b<=0.5,0.05<=c<=0.4,0.01<=d<=0.4,
and 0.15 <=b+c+d<=0.5.
It can be summarized that at the mid 90ties prior art were compositions within the Ni rich
corner of the solid state solution between LiCoO2 - LiMn1/2Ni1/2O2 - {Li1-xNix}NiO2,
including further dopants (like Al). The other corners (UCoO2, in US4302518,
US4357215) and LiN1/2Mn1/2O2 were also known.
During the 90ties there was put little focus on the Li stoichiometry. So the patents above
just claim LiMO2, or a range of Li stoichiometrics, but it has generally not been
understood that the Li:M ratio is an important variable needing optimization. Li1M' was
typically seen as a desired stoichiometry which only can be obtained if a small lithium
excess is used.
In the late 90ties slowly understanding of the role of excess Lithium evolved. The first
document which conclusively shows that additional lithium can be doped into LiMO2 is
JP2000-200607, claiming Li[Co1-xMx]O2 and Li[Ni1-xMx]O2 where M is at least 2 metals

which have an average valence state of 3. Metals M include lithium, Mn, Co, Ni. Not
surprisingly, within the next years several more publications regarding lithium rich
(=Li[L1-xM'-x]O2) materials were published. To our knowledge, the first disclosure of the
possibility of excess lithium, doped into the crystal structure of LiMO2 (M=Mn, Ni, Co)
was JP11-307097, claiming Li(1-a)Ni1-b-c-dMnbCocHdO2 where-0.15 l) is desired to
obtain high rate capabilities.
Another issue is doping to alter the cathode materials. Above mentioned JP3561607
claims lithium nickel-cobalt-manganese oxide doped with at least 1% of a further dopant,
chosen from Al, B, Si, Fe, V, Cr, Cu, Zn, Ga, and W. The patent does not show or explain
why these particular dopants were chosen. JP3141858 disclosed fluorine doped cathode
materials, whereas JP3355102 discloses doped (Mn, Co, B, Al, P, Mg or Ti) LiNiO2 with
a BET surface area of 0.01 - 0.5 m2/g, containing less than 0.5 % SO4.
Anomer issue is the shape of X-ray diffraction peaks. Sharp peaks with narrow FWHM
(full width at half maximum) are related to high crystallinity. JP3653409 (Sanyo) claims a
doped LiNiO2 with FWHM of the main peak at 003 of 0.15-0.22 deg of 2 theta, using Cu
- K alpha radiations.

JP3301931 (Sanyo) claims a doped (> 1%) LiNi-Mn-Co oxide where the main 003 peak
(at 18.71 ± 0.25) has a FWHM < 0.22 degree.
Despite of the impressive numbers of prior art - it is still not fully clear which
compositions within the ternary triangle LiNiO2 - LiCoCO2 - LiNi1/2Mn1/2O2 -
Li[Li1/3Mn2/3]O2 gives the best performance in terms of capacity and rate performance.
The overall development of cathode materials involves improving parameters which
matter in the batteries. Some of the parameters are relatively easy to measure, like
capacity, voltage profile and rate performance, which can be measured by making and
testing coin cells. Other parameters are less obvious. So it is not fully clear how safety or
swelling properties (e.g. of charged polymer batteries during storage at elevated
temperature) can be measured, without assembling real batteries. There exists a strong
indication that these safety and storage parameters are not only determined by the
chemical composition of the cathode but also by surface properties. However, reliable
previous art in this area is rare.
In this respect, the authors observed a problem that resides in the reaction of the surface of
the active lithium transition metal oxide cathode material and the electrolyte in the battery,
leading to poor storage properties and a decreased safety of the battery. The authors argue
that lithium located near to the surface thermodynamically is less stable and goes into
solution, but lithium in the bulk is thermodynamically stable and cannot go to dissolution.
Thus a grathent of Li stability exists, between lower stability at the surface and higher
stability in the bulk. By determining the "soluble base" content, based on the ion exchange
reaction (LiMO2 + δ H+ ↔ L1-δHδMO2 + 8 Li+), the Li grathent can be established. The
extent of this reaction is a surface property.
To improve safety, aluminum doping of LiNiO2 based cathodes, as well as Al, Mg-Ti or
Ni-Ti doping of LiCoO2 has been frequently disclosed, for example in JP2002-151154
(Al+Co doped LiNiO2) or JP2000-200607 (doped LiCoO2). Typical for doping is that the
doped element fits to the host crystal structure, which limits doping of LiMO2 more or less
to transition metals, Li, Mg, Ti, Al, and maybe B. Several disclosures show anionic

doping, like fluorine doping, phosphor doping or sulfur doping. It is however very
questionable if these anions can replace oxygen because they differ in significantly in size
or valence. It is more likely that they instead are present at the surface and grain
boundaries as lithium salts. The lithium salts LiF, Li3PO4 and Li2SO4 all have high
thermal stability which promotes a thermodynamic co-existence with the LiMO2 phase.
In general doping is the modification of the bulk structure, whereas, for safety and storage
properties, the surface chemistry is more important. Unfortunately, in many cases, the
improvement of surface properties is more than outweighed by the deterioration of bulk
properties. Typical examples are the doping by aluminum, where better thermal stability
often is accompanied by a dramatic decrease of power (rate performance).
An alternative approach, widely disclosed in the literature is coating. An early disclosure
of a coated cathode was KR20010002784, where a LiMO2 cathode (M=Ni1-xCox) (or the
sulfur or fluorine "doped" LiMO2 cathode is coated with a metal oxide with metal selected
from Al, AJ, Mg, Sr, La, Ce, V and Ti and the stoichiometric amount of metal is at least
1%.
An alternative approach is the creation of core-shell cathode materials, or grathent type -
cathode materials. Here a thick and dense shell of a more robust cathode material protects
a core of a more sensitive cathode material. Depending on sintering temperature and
chemical composition, the final cathode has either a core-shell morphology or a grathent
morphology. Typically both the shell and the core are electrochemically active (have
reversible capacity). Examples are found in US2006105239 Al, US2007122705 Al or
US2002192552A1.
Sulphate is an impurity of concern in layered lithium transition metal oxides. Sulphate
typically originates from the mixed hydroxide precursors. This is because the mixed
hydroxide preferably is precipitated from transition metal sulphate solution, which is the
cheapest water soluble transition metal precursor. Complete removal of sulfur is difficult
and increases the cost of the precursor. The sulphate impurity is suspected to cause (a)
poor overcharge stability and (b) contribute to the highly undesired low Open Circuit
Voltage (OCV) phenomena, where a certain fraction of batteries show a slow deterioration

of OCV after initial charge. Sulphate impurities normally measured when using transition
metal sulphate solutions in the manufacturing process can be up to 5 wt%.
Finally, manufacturers ate frequently confronted with the presence of very fine particles in
the cathode materials. This is highly undesired because very fine particles - in the final
battery - might electromigrate across the separator, depositing on the anode and causing so-
called "soft shorts". These soft short are highly undesired because they might cause field
failure of batteries.
It is an object of this invention to develop lithium transition metal oxide cathode materials
having improved electrochemical properties, like capacity, voltage profile and rate
performance; besides offering solutions to safety and storage problems that are not only
determined by the chemical composition of the cathode but also by surface properties. Also
the presence of "soft-shorts" oan be eliminated.
The invention discloses a powderous lithium transition metal oxide having a layered
crystal structure Li1+aM'-aO2±b M'k Sm with -0.03 < a < 0,06, b = 0,
M being a transition metal compound, consisting of at least 95% of either one or more
elements of the group Ni, Mn, Co and Ti;
M' being present on the surface of the powderous oxide, and consisting of either one
or more elements of the group Ca, Sr, Y, La, Ce and Zt, with 0.0250 < k ≤ 0.1 in wt% aud
0.15 < m ≤ 0.6, m being expressed in mol%. Preferably 0.25 ≤ m ≤ 0.6.
Preferably also M is consisting of at least 99% of either one or more elements of the group
Ni, Mn, Co, Al, Mg and Ti. Preferably M' is Ca, with 0.0250 ≤ k< 0.0500, and preferably
k ≤ 0.0400.

In another preferred embodiment M= NixMnyCoz with 0.1≤x <0.7,0.1≤y <0.7, 0.1≤z <0.7,
and x+y+z=1. In a special embodiment 1.0 ≤x/y ≤1.3 and 0.1 < z < 0.4, and M comprises
10-15 at.% of Ni2+, and preferably 11.5 -13.5 at.% per total metal Li1+2M'-2.
For M, most preferred is x=y=z=0.33.
This invention demonstrates that the surface properties, determining the safety and
stability of cathodes in real batteries - the surface properties being measured as base
content by pH titration -are strongly determined by the sulfur and the content of elements
like Ca, Sr, Y, La, Ce and Zr.
At least 150 ppm M' (preferably Ca, Sr, Y, La, Ce and Zr) is needed to achieve the
beneficial effect, if the M' addition level is too high (> 1500 ppm), the electrochemical
properties suffer, particularly the rate performance decreases and the irreversible capacity
increases. In a preferred embodiment sulfur levels of 0.15 - 0.6 mol% can be tolerated if
150-1500 ppm of Ca impurity is present. It was found that 0.15-0.6 mol% of sulfur is
harmful to the cathode performance if the Ca doping is lower than 150 ppm.
It is not known and has not been published that Li-Ni-Mn-Co cathode materials, over a
wide stoichiometric range, show a better performance if they contain a certain
concentration of divalent nickel. There is no prior art that teaches that there exists an
optimum Li:M stoichiometric ratio, corresponding to a content of 11.5 -13.5 % of divalent
nickel per metal in the cathode. The actual invention discloses that, surprisingly, the
requirement of 11.5-13.5% of divalent nickel relates lithium excess and Ni:Mn ratio in a
simple manner. This involves that in some cases, surprisingly, a certain lithium deficiency
is preferred.
The invention also covers an electrochemical cell comprising a cathode comprising as
active material the powderous lithium transition metal oxide as described above.
The lithium transition metal oxide can be prepared by a cheap process, for example by a
single firing of a mixture of a suitable precursor and lithium carbonate in air. Preferably

the precursor is a mixed metal precursor like mixed hydroxide, oxyhydroxide or carbonate,
already containing adequate amounts of sulfur and calcium. Hence, the invention further
covers a method for preparing the powderous lithium transition metal oxide described
above, comprising the steps of:
- providing for a mixture of M-sulphaie, a precipitation agent, preferably NaOH or
Na2CO3, and a compiexing agent, hereby
- precipitating a M-hydroxide, -oxyhydroxide or -carbonate precursor from said mixture
having a given sulfur content,
- ageing said precursor whilst adding a base, thereby obtaining a certain base:prccursor
ratio, followed by washing with water, and drying,
- mixing said aged M-hydroxide or -oxyhydroxide precursor with a Li precursor,
- sintering said mixture at a temperature T of at least 900°C, and preferably at least 950°C,
for a time t between 1 and 48 hrs, thereby obtaining a sintered product; where either:
- a salt of M' is added to said M-sulphate containing mixture, or
- M' is added to said base during ageing, or
-M' is added to the water used in said washing step, or
- a M' salt solution is added to a slurry prepared by suspending said sintered product in
water, followed by drying.
Where a salt of M' is added to the M-sulphate containing mixture, this can be to the M-
sulphate itself, to the hydroxide (NaOH) or the compiexing agent.
In the method, preferably M'=Ca and the salt is either one of Ca(NO3)2 and CaCl2.
It is preferred that the sulfur content is controlled during the ageing step by selecting a
given base:precursor ratio.
The actual invention discloses that the application of less than one monolayer of a suitable
element, particularly Ca, dramatically changes the surface properties of layered lithium

transition metal oxides Li1+xM'-xO2, M= Ni-Mn-Co, with -0.03 < x < 0.06. Calcium is a
suitable element but it is very likely that other elements can be added, typical candidates
being rare earths and earth alkali metals, as well as Zr, Pb, Sn.
Surface modified cathode materials are prepared in a single step. Precursors can be
enriched by e.g. Ca to reach a concentration of 150-1500 ppm. These precursors are used
to prepare surface modified LMO by a single cook. If the Ca level of the precursors is
lower, then Ca can be added to the precursor, preferably in liquid form, by a technique
which the authors call slurry doping. High surface area precursor (for example mixed
hydroxide) is dispersed in as little as possible water (or any other solvent) to form a paste
of high viscosity. During rigid stirring a dissolved calcium salt like CaCl2 or Ca(NO3)2 is
slowly added until the desired concentration is reached. During addition, and during the
following drying, calcium precipitates and is well-dispersed onto the surface of the mixed
hydroxide.
Alternatively the calcium can be added during the precursor preparation process. This is
possible by adding a small concentration of calcium (typically less than 100 ppm) to the
water used to dissolve the metal salt (for example MSO4) precursor or base (NaOH)
precursor. Alternatively Ca can be added in higher concentration to the water used to wash
the precursor after finished precipitation.
The surface modification by calcium is possibly a catalytic de-activation of active surface
sites, because (a) Calcium has a much larger ionic radius and cannot be doped into the
bulk structure and (b) up to 1500 ppm Ca is simply not enough to form a coating layer, as
shown below. Here the word coating is used in the conventional sense as a layer
consisting of at least 10-100 atomic layers, corresponding to a few nm to about 100 nm
thickness. The authors speculate that the mechanism of de-activation is related to a
phenomenon known from catalyst technology, called catalyst poisoning. During operation
of a catalyst (for example platinum in a gas containing traces of sulfur species) trace
amounts can de-activate the catalyst by covering catalytically active sites.
The complex layered lithium transition metal oxides are solid state solutions within the
ternary system LiNiO2 - LiCoO2 - LiNi1/2Mn1/2O2 - Li[Li1/3Mn2/3]O2 additionally

including the possibility of lithium deficient cathodes {Li1-xMx}MO2 and not excluding
the possibility of cation mixing
{Li1-xMx}[M(1-y)Liy]O2.
The authors discovered that the optimum Li:M ratio depends on the metal composition.
The authors investigated several metal compositions M=Ni(1-a-b)MnaCob by measuring the
electrochemical performance and base content of test samples as a function of Li.M ratio.
Typically "O" shaped curves (similar as figure 2 of Japanese Patent application 10-
109746) are obtained. The capacity is typically a relatively flat maximum, deteriorating
fast with lower Li:M and more slowly with higher Li:M. The authors discovered that the
maximum (=optimum) of these "O" shaped curves appears at different Li:M ratio's, where
the optimum Li:M ratio depends on the metal composition. Particularly, the optimum
depends on the Ni composition and the Ni:Mn stoichiometric ratio. The authors
discovered that the optimum region is related to the content of divalent nickel as described
below:
M"O2 is a layered ordered rock salt compound with M"=Li1+kM'-k where M contains a
mixture of manganese, cobalt and nickel, -0.03 < k < 0.06. If k>0 then the formula
corresponds to a solid state solution of the ternary system LiNiO2 - LiCoO2 -
LiNi1/2Mn1/2O2 - Li[Li1/3Mn2/3]O2 and can be rewritten as Li[Lix/3Mn2x/3Mny/2Niy/2Co2Ni1.
x-y-z]O2. In this formula all Mn is tetravalent, all cobalt is trivalent and the y/2 Ni is
divalent whereas the 1-x-y-z Ni is trivalent. If k < 0, furthermore assuming that divalent
nickel substitutes for lithium sites, the formula can be rewritten as {Li1-x/3Nix/3}Mny/2.
2x/3Niy/2-x/3CO2:Ni1-x-y-z]O2. In mis formula all Mn is tetravalent, all cobalt is trivalent and the
y/2 Ni is divalent whereas the 1-x-y-z Ni is trivalent.
The authors observed that the optimum Li:M ratio (= (1+x/3) / (1-x/3)) sensitively
depends on the transition metal composition, and corresponding to a quite narrow
stoichiometric range of Ni", which again leads to optimized electrochemical properties. It
is preferred that the Li:M is chosen so that divalent nickel comprises not less than 10%
and not more than 15% of the total metal M" (=Li-M). More preferred, divalent nickel
comprises not less than 11.5 and not more than 13.5 at% of the total metal.

This requirement is strictly valid for layered Li-M-O2 within a certain transition metal
stoichiometric range. The requirement becomes less accurate if the sample is "high Ni",
i.e. Ni1-x-yCoxMny with 1-x-y > 0.6, especially if y < 0.3. The requirement is also less
valid if the sample is "low Ni & low Co", i.e. Ni1-x-yCoxMny with Ni:Mn < 1.3 and x < 0.2.
The requirement of course makes no sense for samples which do not contain enough
nickel, i.e. Ni1-.x-yCoxMny with 1-x-y < 0.2. In the first case (high Ni) there is a trend that
more Ni2+ is required to obtain good electrochemical performance. In the latter case (low
Ni & Co) there is a trend that less Ni2+ is needed. In a medium stoichiometric range Ni1-x.
yCoxMny (i.e. with 0.1 90g of a clear solution which contains the soluble base.
The content of soluble base is titrated by logging the pH profile during addition of 0.1 M
HC1 at a rate of 0.5 ml/min until the pH reaches 3 under stirring. A reference voltage
profile is obtained by titrating suitable mixtures of LiOH and LiCO3 dissolved in low
concentration in DI water. In almost all cases two distinct plateaus are observed. The
upper plateau is OH/H2O followed by CO32-/HCO3-, the lower plateau is HCO3-/H2CO3.

The inflection point between the first and second plateau as well as the inflection point
after the second plateau is obtained from the corresponding minima of the derivative d
pH/d Vol of the pH profile. The second inflection point generally is near to pH 4.7.
Results are listed as micromole of base per g of cathode.
The amount of base which goes into solution is very reproducible, and is directly related
to surface properties of the cathode. Since these have a significant influence on the
stability (i.e. safety and overcharge/high T storage properties of the final battery) there is a
correlation between base content and stability.
Table 1A and 1B summarize the results:

The samples are very similar, with one exception: the soluble base content of sample MP1
(with high Ca) was significantly lower than for MP2. Other properties are very similar,

and although MP2 (with low Ca) shows slightly higher capacity, slightly lower
irreversible capacity and slightly higher rate performance, the results for MP1 are still
acceptable. More important, the samples MP1 and MP2 were sent to battery producer for
safety testing. Whereas MP1 passed the safety test, MP2 did not pass.
The "Safety overcharge test" used here is a safety test where a battery is charged at a very
high rate (for example with 1C charge rate) until a much higher voltage than the normal
operating voltage (for example 20V) is reached. In many cases during such a test more
lithium is extracted from the cathode than can be inserted to the anode, so the
dangerous effect of lithium plating occurs. At the same time the highly delithiated cathode
is in a highly reactive state, and ohmic (resistive) heat is generated. The heat can initiate
the dramatic thermal run-away reaction, ultimately leading to the explosion of the battery.
If a battery passes such a test (i.e. does not explode) or not is strongly dependent on the
choice of cathode material, its morphology, impurity levels and its surface
chemistry. Very little fundamental scientific understanding exists, but the presence of fine
particles definitively contributes to poor safety, (see also below)
Conclusion: the higher content of Ca caused lower soluble base content and higher safety.
Example 1 showed that a Ca content of approx. 250-400 ppm effectively lowered the base
content and improved the safety of the cathode. If we now estimate the number of atomic
layers on top of the surface of the cathode, assuming that
a) all of the calcium is located at the surface of the cathode particles,
b) the surface area of the cathode is reliably obtained by 5 point BET measurement using
nitrogen,
c) Calcium is evenly distributed on the surface,
d) the average distance between Ca atoms is the same as in CaO;

then it can be concluded that the effect of Ca is rather a catalytic effect (less than a few
one atomic layer) and not caused by a conventional coating effect (many layers of atoms).
This is shown in
Example 2: calculation of the "thickness" of the Ca surface layer.
The estimation, based on the data of Example 1, goes as follows:
CaO has an fee crystal structure with 4.8108A lattice constant; thus nearest neighbors
form tetrahedrons with 3.401 A side length. Thus a one-atom monolayer of Ca (having a
hexagonal 2-dim lattice with 3.401A lattice constant) corresponds to a density of 0.664
mg/ m2. The cathode material MP1 (MP2) of Example 1 has a BET area of 0.42 m2/g
(0.44 m2/g). A monolayer covering this BET area corresponds to 280 ppm (292 ppm) Ca.
Therefore sample MP1 has a surface coverage of approx. i.4 monolayers and MP2 has a
coverage of only 0.41 monolayer of calcium. This is much thinner than conventional
coating.
It can be concluded that the observed effect of calcium is not a protection by a coating
layer but rather a catalytic effect (de-activation of active surface sites)
Example 3: theoretical background: Base content / Ca chemistry
It might be argued that a possible dissolution of Ca somehow interferes with the solubility
of lithium or base, thus causing the observation of lower base content for samples with
higher Ca. This argumentation is wrong.
First, Lithium compounds have higher solubility than corresponding Calcium compounds.
Secondly, this example shows that the amount of Calcium is negligible, thus it cannot
change the solubility of Li or base during the pH titration measurement.
We use samples MP1 and MP2 of Example 1 to make the following estimations:
25.9 µmol of base per g of cathode are titrated for sample MP1.
51.2 µmol are titrated for the lower Ca sample MP2.
Thus the content of soluble base differs by 25.2 umol /g.

MPl has 393 ppm Ca, MP2 has 120 ppm Ca. This is a difference of Ca content of 271
ppm.
The molar weight of Ca is 40.1 g/mol.
A simple calculation yields that the difference in Calcium is 271/40.1=6.76 umol /g.
We conclude that an increase of Ca by only 6.76 µmol/g causes a much larger decrease of
base by 25.2 µmol. The large decrease can only be explained if we accept that Ca
stabilizes the surface so that less Li goes into solution.
Example 4: Base content as a function of Ca content - different precursors
Example I demonstrated that the low Ca sample MP2 had higher base content than high
Ca sample MPl. This is confirmed in Example 4 by detecting a good correlation between
lower base content and higher Ca content for a larger series of samples with similar
morphology and composition (Li, Mn, Co, Ni, S).
Ten transition metal hydroxide precursors from a mass production batch were received,
denominated MOOH1-10. The hydroxides have a metal composition of
M=Ni1/3Mn1/3Co1/3. Ten samples - lithium transition metal oxide samples S1a-S10a (each
approx. 250g) - were prepared with a Li:M=l.l blend ratio (according chemical analysis
of the precursor) at a temperature of 960°C in air. The lithium content was checked (by
comparing the unit cell volume) and the BET surface area was measured. All samples had
a very similar morphology (particle size distribution, tap density, particle shape, SEM
micrographs, crystallite size).
The Ca content of all precursors was obtained by chemical analysis. The content of Ca in
the final product is the same as in the precursor. The crucibles do not contain Ca,
evaporation is not observed, and Ca practically does not diffuse into me crucible. The
soluble base content of the Li-M-oxide samples was measured by pH titration.
Tables 2A and Figure 1 summarize the results.

Table 2 A Properties of samples S1a-S10a prepared from transition metal hydroxide
precursors.

Then a second series of test samples S1b-S10b (each approx. 700g) was prepared. The
temperature and Li:M blend ratio was corrected slightly to achieve samples with a more
narrow distribution of BET and identical final Li:M ratio. Table 2B and Figure 1
summarize the results. It contains the data of Table 2A (A: bullets O) and data of some
further samples (mass production samples), indicated as stars (B: *)

Table 2B Properties of samples S1b-S10b prepared from transition metal hydroxide
precursors

Apparently, there exists a clear correlation between increasing Ca content and lower
soluble base content. The example confirms that a small amount of Ca dramatically
decreases the amount of soluble base, without much deteriorating the electrochemical
performance: a slight increase of irreversible capacity and a slight deterioration of rate
performance are observed. As expected, the normal impurity level of Ca (<150 ppm) gives
the worst results for base content. Figure 2 summarizes the measured electrochemical
properties as function of calcium content, taken from Tables 2A and B (indicated by
bullets O). The left figure plots the irreversible capacity (%) vs. Ca content, the right
figure the rate performance at 2C (%) vs. Ca content. Data for irreversible capacity of
some further samples (mass production samples) were added to the Figure as stars (*).

In practice it is worth to accept the slight deterioration of rate performance if this allows to
dramatically lower the base content, thus achieving improved high temperature stability
and safety of real cells.
Example 5: Soluble base content and Electrochemical performance as a function of
effective S-Ca content
Most samples of Example 4 had a similar level of sulfur. Example 5 will show that the
content of Ca and the content of sulfur completely determines the soluble base content as
well as other properties (electrochemically performance) for a larger series of mass scale
production samples ( > 500 kg sample size). The samples had the same composition (Li,
Mn, Ni, Co) but differed in Ca and Sulfur content
Data analysis showed that Ca has a negative regression coefficient versus the soluble base
content, whereas the SO4 content has a positive regression coefficient. This allowed to
define a statistical variable k being the "effective S-Ca" content by k = 0.84 * S- Ca
where S and Ca are the ppm results of the ICP analysis for S and Ca. The formula can be
interpreted as the statistical proof that a higher content of sulfur can be neutralized by
addition of Ca.
Figure 3 shows mat there is a very good correlation between effective S-Ca content and
soluble base content. Both Ca and Sulfur correlate reasonable well with base content. The
top left figure gives the soluble base content (µmol/g) vs. Ca content, the bottom left
figure gives the same against the SO4 content. A statistical variable k (a linear combination
of 0.84 * S (ppm) - Ca (ppm)) shows an almost perfect positive correlation. The
correlation coefficient is + 0.95. This is shown on the right figure.
Surprisingly, there is also a very good correlation between soluble base content (umol/g)
and electrochemical performance, as shown in Figure 4. Here the electrochemical
performance is given by the discharge capacity of the first cycle (1st cycle DC Q - in
mAh/g). The correlation factor is 0.94.

Figure 3 and Figure 4 are important examples showing the need to control very well the
Ca level and S levels. Note that the base content varies by almost 100%, and the discharge
capacity by 5%, these are comparably huge numbers considering that the Ca content
varies by less than 600 ppm and the sulfur content by about 0.25 mol%
Example 6: Optimization of Ca and Sulfur additions.
This Example serves to demonstrate 2 aspects of the invention:
(1) it confirms the observation of Example 5 that Ca "neutralizes" the negative effect of a
high soluble base content of sulfur containing cathodes, and
(2) it demonstrates that only samples which contain both sulfur and calcium according to
the invention show good overall performance.
The Example uses a mixed transition metal hydroxide precursor with metal composition
M=Mn1/3Ni1/3Co1/3. The precursors naturally are low in Ca but contain some sulfur. The
sulfur is removed after preparation of a preliminary Li-M-Oxide sample (Li:M = 1.1) by
washing. Then the preliminary sample is used as precursor, and the following material
matrix is prepared:
(6a): no addition of sulfur or calcium
(6b): addition of 400 ppm Ca
(6c): addition of 0.5 wt% SO4
(6d): addition of both 400 ppm Ca and 0.5 wt% SO4,
This is followed by a re-sintering. Final samples with the same morphology but different
Ca, S composition are obtained. The addition of Ca and S is performed by slurry doping of
the Li-M-oxide preliminary sample (also described below in example 7). Slurry doping is
the drop-wise addition of a Li2SO4 solution or of a Ca(NO3)2 solution during stirring of a
preliminary sample powder-in-water slurry of high viscosity, followed by drying in air. A
total of 400 ppm Ca and/or 5000 ppm (SO4) sulfur was added. Note that 1000 ppm of
sulfate generally corresponds to approx. 0.1 mol% of sulfur, more accurate- for
Li1.04M0.96O21000 ppm correspond to 0.105 mol %.

The experiment was repeated for a precursor with M=Ni0.53Mn0.27Co0.2 composition,
where the preliminary sample - the precursor during slurry doping - was prepared using a
Li:M=1.02 blend ratio. The conclusions of Example 5 (neutralization of sulfur by Ca) are
confirmed: if the sample contains sulfur, the addition of Ca neutralizes the high soluble
base content caused by the sulfur.
Electrochemical properties are tested, and settling down kinetics is measured (see also
Example 8 for more details). The sample without added Ca showed the highly undesired
fine particles which do not settle down. All samples with Ca settled down very fast.
Of all samples - only the sample which contains Ca and sulfur show overall good
performances, as can be seen in Tables 3A and 3B.
Samples situated outside the claimed concentrations (either too high or too low) show the
following disadvantages:
Low Ca & low SO4 -> unacceptable level of fine particles
Low Ca and high SO4 -> high soluble base content, fine particles
High Ca and low SO4 -> relatively poor electrochemical performances.
(see also below Table 4A)



Note that in this test (3B) some of the added SO4 was lost due to crystallization.
Example 7: Comparing Ca and Mg with same precursor material
This example shows data of different samples prepared from one single hydroxide
precursor, with varying Ca concentration by addition of different amounts of Ca to the
precursor during preparation. As reference Mg was added to confirm the role of Ca.
A hydroxide with low content of Ca (60 ppm) was received. The transition metal
composition was approx. Ni0.37Co0.32Mn0.31. Sulfur content was approx. 0.4 wt% SO4. The
hydroxide was divided into smaller samples (each approx. 500g). A water-based slurry of
high viscosity was prepared from each sample. The water used to slurry the precursor
contained appropriate additions of dissolved CaCl2 • The slurry was continuously stirred.
Thus a Ca doped slurry was achieved which was dried in a convection oven without
filtering, resulting in a Ca treated mixed hydroxide. In the same way Mg doped (dissolved
Mg(NO3)2 was added to the water) and Mg + Ca doped mixed hydroxide was prepared
from the same precursor.

Six samples (CaAddI - CaAdd6) were prepared from the Ca doped mixed hydroxide by
mixing with Li2CO3, the Li:M blend ratio was 1.07, followed by a heating at 960°C.
Table 4 gives an overview of the prepared samples. The Ca concentration of the undoped
sample is slightly higher than expected (120 ppm), possibly caused by a slight Ca
dissolution from the baker used during the slurry preparation. The table shows mat (a) the
content of base decreases with increasing content of Ca and (b) the addition of Mg does
not alter the base content at all, (c) the BET surface area decreases with increasing
calcium content, (c) indicates that the sintering kinetics speeds up with higher Ca (or Cl
contamination) content.
The base content decreases by 33%, whereas the BET area only decreases by 18%,
proving that part of the decreased base content is caused by a different surface chemistry,
and not, as could be assumed, by a decrease of the surface area itself. Note that the
reduction of base is slightly less than expected, possibly caused by a less than perfect
dispersion of Ca on the surface of the precursor during slurry doping.
Table 4 also shows that the magnesium does not influence the base content at all. The base
content however depends on the Ca content, independently of how much Mg is added.
The soluble base decreases with increasing calcium level. It is believed that the ionic
radius of Mg is too small (0.66 Angstrom) compared to Ca (0.99 Angstrom), the latter
having a size that fits very well to the surface of Li-M-oxide - see Example 11 below.

Table 4: Properties of samples prepared from a single MOOH modified by adding Ca
and/or Mg

Example 8: Ca level and fine particles
As said above, the presence of very fine particles is highly undesired because very fine
particles - in the final battery - might electromigrate across the separator, depositing on
the anode and causing so-called "soft shorts", leading to field failure of batteries. These
particles are normally finer than 1 µm. It is believed mat the decrease of these fine
particles is responsible for better safety.
Example 8 shows that the addition of Calcium eliminates fine particles, although the
mechanism causing this beneficial effect is not fully understood by the authors.
The samples CaAdd1, CaAdd2, CaAdd3 and CaAdd4 (of Example 7) were investigated in
a settling experiment. After disposing a cathode material in water it is desired that the
particles settle down fast, and that a clear solution remains on top. A slow settling
indicates the presence of fine particles.

Figure 5 shows photographs of a settling down experiment. Ca content: from left to right:
(1) 120 ppm (2) 190 ppm, (3) 420 ppm, (4) 900 ppm. After a settling time of 1 minute of 5
g of suspended particles in a 50 ml measuring (graduated) cylinder, the height of the
separation line between clear solution and the particle suspension layer was situated at (1)
50, (2) 30, (3) 22, and (4) 13 ml, after 5 min: (1) 49, (2) 11, (3) 9, and (4) 8 ml. Obviously,
an increase of Ca impurity causes a dramatic increase of settling kinetics - proving that Ca
addition eliminates the presence of fine particles.
As a result of Examples 4 to 8 the following Table 4A gives an overview of the addition
of Ca and S.


Example 9: Soluble base content is a thermodynamic materials property
This example discusses that the base content is a thermodynamic equilibrium materials
property. It can be changed by well-designing the calcium and sulfur content. It can,
however not be changed by altering the preparation conditions after being sufficiently
equilibrated.
A few kg of the mass production sample MP2 of Example 1 (having a "natural" Ca
content of 120 ppm) is used to investigate if the soluble base content can be lowered,
depending on heating temperature, air flow, or washing in water followed by reheating.
For the washing - the amount of water is limited, and the Li lost is monitored. This figure
is negligible, consisting of approx. 0.1% of the total Li in the sample. Reheating
temperatures are lower than the initial sintering temperature, thus the morphology does
not change during reheating.
The soluble base content of the initially received sample can be slightly lowered by a heat
treatment (equilibration), indicating that the lithiation of the MP2 sample is not 100%
completed. However, after reheating, independently of heating conditions, the same
soluble base content is always achieved. This base content is the equilibrium content,
depending of surface area, metal composition and Ca and sulfur level. Washing removes a
large fraction of Sulfur - as soluble Li2SO4 - but does not remove Ca (this was checked by
ICP), resulting in a low Sulfur - low Ca sample. The low sulfur - low calcium sample has
a lower soluble base content. After washing, already at low drying temperature (150°C)
the same equilibrium value is re-established which is achieved after washing and reheating
at 750°C. All these observations are summarized in Table 5.

Table 5:

Example 10: Comparison of identical morphology with high/ low Ca content
A sample EX10A (1 kg size) is prepared from a mass scale production precursor mixed
hydroxide with metal composition Mn1/3Ni1/3Co1/3 by mixing the precursor with Li2CO3
(blend ratio 1.1) followed by heating to 960°C. EX10B is prepared in the same way, with
the exception that the precursor was modified by the previously described slurry doping: a
total of 400 ppm Ca was slowly (drop wise) added in the form of Ca(NO3)2 to a water
based slurry of the precursor, followed by drying (no filtering).

Table 6A and 6B summarize the results

As Tables 6A and 6B show, besides of the Ca impurity level, all 3 samples are, as
expected for samples prepared under similar conditions from the same precursor, very
similar. The PSD, obtained by laser diffraction are identical. Similar as observed in
previous examples - the sample with Ca addition shows the smallest content of soluble
base.
Despite that the particle size distribution of sample EX10A and EX IOC is identical - the
settling down kinetics after dispersing the cathode in water is dramatically different.
Figure 6 shows photographs of a settling down experiment of Ca treated LiM02 :
Addition of Ca: (1: left) 0 ppm, (2: right) 400 ppm: after a settling time of 1 minute of 5 g
of suspended particles in a 50 ml measuring (graduated) cylinder, the height of the
separation line between "clear" solution and the particle suspension layer was situated at
(1) 27, (2) 18 ml, only after 5 min the suspended particles in both cylinders have nearly all

settled down. Obviously, an increase of Ca impurity causes a dramatic decrease of fine
particles - as a result Ca rich cathodes settle down much faster.
Example 11: Alternative elements besides Ca.
This example uses a mixed transition metal hydroxide precursor with metal composition
M'=Mn0.33Ni0.38 C00.29 as precursor. The precursor is low in Ca but, as expected, contains
some sulfur. A similar experiment is done with a mixed hydroxide precursor with
M2=Ni0.53Ni0.27Co0.2 composition.
The precursors are doped by slurry doping: 1000 ppm of nitrate solutions of Ca, Y, Sr, La,
Ba, Fe are added, respectively. A reference was slurry doped but no metal was added.
After slurry doping the precursors were mixed with Li2CO3 and cooked. Besides of the
doping, final composition (Li, Mn, Ni, Co) was very similar.
To compare the efficiency to iower the base content the following parameters are
considered:
(a) Soluble base content (= soluble base / mass of cathode)
(b) Specific surface base (= soluble base content / surface area of cathode)
(c) Molar efficiency of dopant (µmol) versus gravimetric efficiency of dopant
(ppm)
The results are summarized in Tables 7A (M') and 7B (M2) below.

The conclusions are as follows:
(a) Base content: Sr and Ca, and to a lesser degree Y and Ba are most efficient to lower
the soluble base content
(b) The final samples have different BET area, hence the "Specific Surface Base Content"
is observed: Ca, Sr and Y, and to a lesser degree La lower the specific surface base content
of the cathode.
(c) Gravimetric efficiency: Sr and Ca are the most efficient. Molar efficiency: Considering
the high molecular weight of Y (more than twice that of Ca) we conclude that both Ca and
Y are most efficient to neutralize high base caused by sulfur. Sr is somewhat less effective
and La shows noticeable, but small efficiency. Ba is not effective, as can be seen in the
"Specific Surface Base Content". Fe is inert (not reported).
The authors speculate that the effective elements have an ionic radius of 0.7 to 1.2
Angstrom. Especially Ca and Y - which have almost similar and quite small ionic radius
(in 6 coordination Ca: 0.99, Y: 0.893 A) - have a size that fits very well to the surface of
Li-M-oxide. The more preferred range for ionic radii is 0.85-1.15 Angstrom.
Example 12: Strontium versus Calcium
Example 11 compared the efficiency of Ca, Sr, La, Ba, Y to lower the content of soluble
base.
However, Example 11 did not take into account that the sintering kinetics change with
different additives - yielding very different BET values. Example 12 compares the effect
of Ca and Sr more carefully.
A reference without addition of additive (Ca or Sr) was prepared from a mixture of mixed
transition metal hydroxide (M=Ni0.38Mn0.33Co0.28) and Li2Co3 at 980°C. Further samples
with addition of 400 and 1000 ppm Sr and 400 ppm Ca were prepared. Each sample used
1 kg of MOOH + Li2CO3. The additive (Ca, Sr) was added by the previously described

"slurry doping" process. Appropriate amounts of solution of Sr(NO3)2 and Ca(NO3)2 were
added to a high viscous slurry of the precursor hydroxide during rigid stirring.
The sintering temperature was adjusted to achieve a similar sintering. Base content was
measured, unit cell volume and crystallite size was obtained from X-ray diffraction and
electrochemical properties were tested by coin cells. Table 8A and 8B summarizes the
preparation conditions results
Table 8 A: Preparation and morphology of samples with Sr, Ca addition

Table 8B Electrochemical performance (capacity, irreversible capacity and rate (versus
0.1C) of samples with Sr, Ca addition

The morphology (BET, particle size) of all samples was basically identical. Ca addition is
most effective to lower the base content. 1000 ppm Sr reduces the base content about the
same, but less than for 400 ppm Ca. However, Sr is interesting because it reduces the base

and at the same time the electrochemical properties deteriorate less than for 400 ppm Ca
addition.
Example 13: What is the optimum Li-Mn-Ni-Co composition ?
So far, this invention demonstrated that the surface properties, determining the safety and
stability of cathodes in real batteries - the surface properties being measured as base
content by pH titration -are strongly determined by the sulfur and Ca (amongst others)
content. The authors also analyzed large amounts of data to understand what else
determines the base content. The analysis shows clearly that the base content furthermore
depends on BET surface area of Li-M-02, it also varies strong with Li:M ratio and Ni:Mn
ratio.
The base content increases linearly with BET, it increases with increasing Li:M ratio and
with increasing Ni:Mn ratio. Table 9 shows a typical example for Li-M-oxidc where M
contains 33% Co.

Table 9: Base content as function of Ni:Mn ratio and Li:M ratio for samples prepared at
different temperature

The authors intended to optimize the BET and Li:M as well as Ni:Mn composition in
order to achieve the optimum of high electrochemical performance, but keeping base
content low. It was shown that a similar electrochemical performance can be achieved by
high BET but lower Li:M, or lower BET and higher Li:M. By trying to optimize the
composition, BET and crystallinity - it was recognized that within the region of interest
only samples with a certain content of divalent Ni, high crystallinity allows to achieve
overall optimized cathodes.
Table 10 below lists the preferred upper and lower Li:M stoichiometric range for Li-M-O2
with different transition metal composition. The columns in the table refer to the following
formulas
(a) Li1+kM'-kO2 with Ni1-a-bMnaCob and (b) Li[Lix/3Mn2x/3Niy/2Mn2/yCozNi1-x-y.]O2 as
follows:

Ni, Mn, Co are the mol fractions 1-a-b, a, b in the transition metal M
"Ni:Mn" is the molar ratio of Ni to Mn (=(1-a-b)/a) in the transition metal M
"Li:M" is the molar ratio of Li:M (=(1+k)/(1-k) = (1+x/3)/(1-x/3)
The column "Ni2+M gives twice the fraction of divalent nickel (= 2 * y/4).

Table 10: Preferred upper and lower Li:M stoichiometric range for Li-M-O2 with different
transition metal composition.

An analysis of the data reveals:
(1) It is difficult to obtain a good overall performance if Ni:Mn=l. Ni:Mn > 1 allows for

better electrochemical performance.
(2) The optimum Li:M stoichiometric region depends on the transition metal composition.
The optimum Li:M is achieved if the cathode Li1+aM'-aO2 contains 11.5-13.5% of
divalent nickel per 2 mol metal (Li+M).
The optimum Li:M decreases with increasing Ni:M.
(a) Ni:Mn=0.95 : Li:M = 1.07
(b)Ni:Mn=1.05 :Li:M=1.06
(c)Ni:Mn=1.2 :Li:M=1.05
(d)Ni:Mn=1.3 :Li:M=1.04
Similar experiments were repeated for different metal compositions, including

WE CLAIM:
1. A powderous lithium transition metal oxide having a layered crystal structure
Li1+aM'-aO2±b M'k Sm with -0.03 < a < 0.06, b = 0,
M being a transition metal compound, consisting of at least 95% of either one or more elements
of the group Ni, Mn, Co and Ti;
M' being present on the surface of the powderous oxide, and consisting of either one
or more elements of the group Ca, Sr, Y, La, Ce and Zr, with 0.0250 < k ≤0.1 in wt%;
and 0.15 < m ≤0.6, m being expressed in mol%.
2. A powderous lithium transition metal oxide according to claim 1, characterized in that M Is
consisting of at least 99% of either one or more elements of the group Ni, Mn, Co, Al, Mg and Ti.
3. A powderous lithium transition metal oxide according to claims 1 or 2, characterized in that M'
is Ca, with 0.0250 ≤ k < 0.0500, and preferably k ≤0.0400, in wt%.
4. A powderous lithium transition metal oxide according to any one of claims 1 to 3,
characterized in that 0.25 ≤ m ≤ 0.6, in mol%.
5. A powderous lithium transition metal oxide according to any one of claims 1 to 4,
characterized in that M= NixMnyCoz with 0.1≤x ≤0.7, 0.1≤y ≤0.7, 0.1≤z ≤0.7, and x+y+z-1.
6. A powderous lithium transition metal oxide according to claim 5, characterized in that
1.0 ≤x/y ≤1.3 and 0.1 < z < 0.4, and comprising 10-15 at.% of Ni2*, and preferably 11.5 -13.5 at.%
per total metal Li1+aM'-a.
7. A powderous lithium transition metal oxide according to claim 6, characterized In that
x=y=z=0.33.
8. An electrochemical cell comprising a cathode comprising as active material the powderous
lithium transition metal oxide according to any one of claims 1 to 7.
9. A method for preparing a powderous lithium transition metal oxide according to any one of
claims 1 to 8, comprising the steps of:
- providing for a mixture of M-sulphate, a precipitation agent, preferably NaOH or Na2CO3, and a
complexing agent, hereby

- precipitating a M-hydroxide, -oxyhydroxide or -carbonate precursor from said mixture having a
given sulfur content,
- ageing said precursor whilst adding a base, thereby obtaining a certain base:precursor ratio,
followed by washing with water, and drying,
- mixing said aged M-hydroxide or -oxyhydroxide precursor with a Li precursor,
• sintering said mixture at a temperature T of at least 900oC, and preferably at least 950oC, for
a time t between 1 and 48 hrs, thereby obtaining a sintered product;
characterized in that either:
- a salt of M' is added to said M-sulphate containing mixture, or
- M' is added to said base during ageing, or
- M' is added to the water used in said washing step, or
- a M' salt solution is added to a slurry prepared by suspending said sintered product in water,
followed by drying.
10. A method for preparing a powderous lithium transition metal oxide according to claim 9,
wherein M'=Ca and said salt is either one of Ca(NO3)2 and CaCl2.
11. A method for preparing a powderous lithium transition metal oxide according to claims 9 or
10, wherein said given sulfur content is controlled during said ageing step by selecting a given
base: precursor ratio.

The invention covers a powderous lithium transition metal oxide having a layered crystal structure Li1+a-M1-aO2±b M'k
Sm with -0.03 < a < 0.06, b=0, M being a transition metal compound, consisting of at least 95% of either one or more elements of
the group Ni,Mn, Co and Ti; M' being present on the surface of the powderous oxide, and consisting of either one or more elements
from (IUPAC) of the Periodic Table, each of said Group 2,3, or 4 elements having an ionic radius between 0.7 and 1.2 Angstrom. M'
however not comprising Ti, with 0.015 < k < 0.15,k being expressed in wt%, and 0.15 < m ≤0.6, m being expressed in mol%. The
addition M' (like Y, Sr, Ca, Zr,...) improves the performance as cathode in rechargeable lithium batteries. In a preferred embodiment
a content of 250-400 ppm calcium and 0.2-0.6 mol% of sulfur is used. Particularly, a significantly lower content of soluble base
and a dramatically reduced content of fine particles are achieved. Especially preferred performance is achieved if 11.5-13.5% of the
metal atoms of the cathodes are divalent nickel.