Double-shell core lithium nickel manganese cobalt oxides. TECHNICAL FIELD AND BACKGROUND The invention relates to cathode material for rechargeable lithium batteries, particularly lithium nickel manganese cobalt oxides being coated with a fluorine containing polymer and heat treated afterwards. Previously LiCoO2 was the most used cathode material for rechargeable lithium batteries. However, recently a substitution of LiCoO2 by lithium nickel oxide based cathodes and by lithium nickel manganese cobalt oxides is in full progress. In these substitute materials, depending on the choice of metal composition, different limitations occur or challenges need to be solved. For simplicity reasons, the term "Lithium nickel oxide based cathodes" will be further referred to as "LNO", and "lithium nickel manganese cobalt oxides" will be further referred to as "LMNCO". One example of an LNO material is LiNi0.80CO0.15Al0.05O2 . It has a high capacity, however it is difficult to prepare, since typically a carbon dioxide free atmosphere (oxygen) is needed and special carbonate free precursors like lithium hydroxide are used instead of lithium carbonate. Hence such manufacturing restraints tend to increase the cost of this material considerably. LNO is a very sensitive cathode material. It is not fully stable in air, which makes large scale battery production more difficult, and - caused by its lower thermodynamic stability - in real batteries it is responsible for a poor safety record. Finally, it is very difficult to produce lithium nickel oxide with a low content of soluble base. By "soluble base" is meant lithium located near to the surface that is less stable thermodynamic ally and goes into solution, whilst lithium in the bulk is thermodynamically stable and cannot be dissolved. Thus a gradient of Li stability exists, between lower stability at the surface and higher stability in the bulk. The presence of "soluble base" is a disadvantage because a high base content is often connected with problems during battery manufacturing: during slurry making and coating high base causes a degradation of the slurry (slurry instability, gelation) and high base is also a responsible for poor high temperature properties, like excessive gas generation (swelling of the batteries) during high temperature exposure. By determining the "soluble base" content by pH titration, based on the ion exchange reaction (LiMO2 + 5 H+ ? Li1.dHdMO2 + d Li+), the Li gradient can be established. The extent of this reaction is a surface property. In US2009/0726810A1 the problem of soluble base is further discussed; LiMO2 cathode material is prepared using mixed transition metal hydroxides as precursors. These are obtained by co- precipitating transition metal sulphates and technical grade bases like NaOH, which is the cheapest industrial route for LiMO2 precursor preparation. This base contains CO32- anion in the form of Na2CO3, which is trapped in the mixed hydroxide - the mixed hydroxide typically containing between 0.1 and 1 wt% of CO32". Besides the transition metal precursor, the lithium precursor Li2CO3, or a technical grade LiOH*H20, containing at least 1wt% of U2CO3 is used. When the lithium and transition metal precursors are reacted at high temperature, typically above 700oC. In the case of high nickel cathode LNO, the. Li2CO3 impurity remains in the resulting lithium transition metal oxide powder, especially on its surface. When higher purity materials are used, less Li2CO3 impurity is found, but there is always some LiOH impurity that reacts with CO2 in the air to form Li2CO3. Such a solution is proposed in JP2003-142093, however the use of expensive precursors of very high purity is not preferred. An example of LMNCO is the well known Li1+xM1-xO2 with M=Mn1/3Ni1/3Co1/3O2 , where the - manganese and nickel content is about the same. "LMNCO" cathodes are very robust, easy to prepare, have a relatively low content of cobalt and thus generally tend to cost less. Their main drawback is a relatively low reversible capacity. Typically, between 4.3 and 3.0V the capacity is less than or about 160 mAh/g, compared with 185-195 mAh/g for LNO cathodes. A further drawback of LMNCO compared with LNO is the relatively low crystallographic density, so the volumetric capacity is also less; and a relatively low electronic conductivity. In between LNO and LMNCO type materials we can situate "Nickel rich lithium nickel manganese cobalt oxides" Li1+xM1-xO2 where M=Ni1-x.yMnxCOy or M=Ni1-x-y-zMnxCoyAlz , with Ni: Mn larger than 1, having typically values for Ni:Mn of 1.5 to 3, and a Co content "y" typically between 0.1 and 0.3. For simplicity we refer to this class of materials as "LNMO". Examples areM=Ni0.5Mri0.3CO0.2, M=Ni0.67Mn0.22CO0.11, and M=Ni0.6Mn0.2CO0.2. Compared with LNO, LNMO can be prepared by standard processes (using a Li2CO3 precursor) and no special gas (such as oxygen as mentioned above) is needed. Compared to LMNCO, LNMO has a much higher intrinsic capacity and possibly a lower tendency to react with electrolyte (which is normally characterized by dissolution of Mn) at elevated temperature. Thus it becomes apparent that LNMO will possibly play a major role in the substitution of LiCo02. Generally, the base content increases, and the safety performance tends to deteriorate with increasing Ni:Mn ratio. On the other hand it is widely accepted that high Mn content helps to improve safety. A high base content is related to moisture sensitivity. In this regard LNMO is less moisture sensitive than LNO but more sensitive than LMNCO. Directly after preparation, a well prepared LNMO sample has a relatively low content of surface base, and if it is well prepared most of the surface base is not Li2CO3 type base. However, in the presence of moisture, airborn CO2 or organic radicals reacts with LiOH type base to form Li2CO3 type base. Similar, the consumed LiOH is slowly re-created by Li from the bulk, thus increasing the total base (total base= mol of Li2CO3 + LiOH type base). At the same time, the moisture (ppm H20) increases. These processes are very bad for battery making. Li2CO3 and moisture are known to cause severe swelling, and to deteriorate the slurry stability. Hence it is desired to decrease the moisture sensitivity of LNMO and LNO materials. In US2009/0194747A1 a method to improve the environmental stability of LNO cathode materials is described. The patent discloses a polymer coating of nickel based cathode materials, in the form of a single layer of non-decomposed polymer. The polymers (e.g. PVDF) are chosen from binders typically used in the manufacturing (slurry making for electrode coating) of lithium ion batteries. Thermal stability (safety) is related to interfacial stability between electrolyte and cathode material. A typical approach to improve the surface stability is by coating. Many different examples of coatings are available in literature and especially in patent literature. There are different ways to categorize coatings. For example, we can distinguish between ex-situ and in- situ coating. In ex-situ coating a layer is coated onto the particles. The coating can be obtained by dry or wet coating. Generally the coating is applied in a separate process involving at least the coating step and generally an additional heating step. Thus the total cost of the process is high. Alternatively, in some cases an in-situ coating - or self organized coating - is possible. In this case the coating material is added to the blend before cooking, and during cooking separate phases form, preferable the coating phase becomes liquid, and if the wetting between LiMO2 and the coating phase is strong then a thin and dense coating phase ultimately covers the electrochemical active LiMO2 phase. Evidently, in-situ coating is only efficient if the coating phase wets the core. We can also distinguish between cationic and anionic coating. An examples for cationic coating is Al2O3 coating. Examples for anionic coating are fluoride, phosphate, silicate coating and the like. Fluoride coating is especially preferred because a protecting film of LiF is formed. Thermodynamically LiF is very stable, and does not react with electrolyte, thus LiF coating is very promising to achieve a good stability at high temperature and voltage. A typical method, such as used by Croguennec et al. in Journal of The Electrochemical Society, 156 (5) A349-A355 (2009), is the addition of LiF to the lithium transition metal oxide to achieve the protecting LiF film. However, due to the high melting point of LiF and also due to poor wetting properties, it is not possible to obtain a thin and dense LiF film. Croguennec reports that, instead of a coating, small particles or 'sheets' can be found in the grain boundaries of the LiMO2 particles. Further possible methods are the use of MgF2, AIF3 or lithium cryolite. We can further distinguish between inorganic and organic coating. An example of organic coating is a polymer coating. One advantage of polymer coating is the possibility of obtaining an elastic coating. On the other hand, problems arise from poor electronic conductivity, and sometimes the poor transport of lithium across the polymer. Generally, polymer coating more or less adheres to the surface, but it does not chemically change the surface. There cannot be found any experimental data in the prior art that would show that the above described approaches are effective to improve the cited problems of LNO and LNMO materials. To summarize: 1) LMNCO is a robust material but has severe capacity limitations, 2) It is desired to increase the thermal stability and to reduce the base content of LNO, 3) It is desired to increase the thermal stability and reduce the base content of LNMO. It is an aim of the present invention to improve or even overcome the problems cited before, and to provide for high capacity alternatives for LMNCO materials. SUMMARY Viewed from a first aspect, the invention can provide a lithium transition metal oxide powder for use in a rechargeable battery, having the surface of the primary particles of said powder coated with a first inner and a second outer layer, the second outer layer comprising a fluorine- containing polymer, and the first inner layer consisting of a reaction product of the fluorine- containing polymer and the primary particle surface. In one embodiment this reaction product is LiF, and the lithium originates from the primary particles surface. In another embodiment, the fluorine in the reaction product LiF originates from partially decomposed fluorine- containing polymer present in the outer layer. In a particular embodiment the first inner layer consists of a LiF film with a thickness of at least 0.5 nm. or at least 0.8 nm, or even at least 1 nm. In another particular embodiment, the fluorine-containing polymer is either one of PVDF, PVDF-HFP or PTFE. The fluorine-containing polymer can be composed of agglomerated primary particles having an average particle size of between 0.2 and 0.5 urn. It is believed that such a particle size is advantageous for the wetting properties of the molten fluorine-containing polymer. An example of the lithium transition metal oxide can be either one of: - LiCodMeO2, wherein M is either one or both of Mg and Ti, with e<0.02 and d+e=1; - Li1+aM'1-a02±b M1kSm with -0.030, b">0, c">0 and a"+b"+c"=1; and a"/b">1. In another embodiment 0.51 are particularly suitable for use in lithium-ion prismatic or polymer batteries. The pristine polymer applied for the initial coating contains fluorine. In one embodiment it contains at least 50% by weight of fluorine. A typical example of a pristine polymer is a PVDF homopolymer or PVDF copolymer (such as HYLAR ® or SOLEF® PVDF, both from Solvay SA, Belgium). Another known PVDF based copolymer is for example a PVDF-HFP (hexa-fluoro propylene). Such polymers are often known under the name "Kynar®". Teflon, or PTFE, could also be used as polymer. Viewed from a second aspect, the invention can provide a process for covering a lithium transition metal oxide powder with a fluorine containing double-layered coating, comprising the steps of: providing a bare lithium transition metal oxide powder, mixing this powder with a fluorine-containing polymer, and heating the obtained powder-polymer mixture at a temperature between at least 50° C and at most 140oC above the melting temperature of the fluorine-containing polymer, whereby, on the surface of the metal oxide powder a double-layered coating is formed, consisting of an outer layer consisting of fluorine-containing polymer, and an inner layer consisting of a reaction product of the powder surface and the polymer. In one embodiment, the amount of fluorine-containing polymer in the powder-polymer mixture is between 0.1 and 2 wt%, in another embodiment between 0.2 and 1 wt%. Also, the inner layer preferably consists of LiF. An example inner layer has a thickness of at least 0.5 nm, or at least 0.8 nm, and even at least 1 nm. One example process uses a fluorine-containing polymer such a PVDF, and the powder-polymer mixture is heated at a temperature between 220 and 325°C for at least one hour. In a particular embodiment, heating is between 240 and 275oC for at least one hour. An example of the lithium transition metal oxide used in the process is either one of: - LiCodMeO2, wherein M is either one or both of Mg and Ti, with e<0.02 and d+e=1; - Li1+aM'1+aO2±b M1k Sm with -0.030, b">0, c">0 and a"+b"+c"=1; and a"/b">1. In another embodiment 0.5=a"=0.7, 0.1 175°C, and a stepwise increase at T = 300 - 325 °C. The increase of lattice constants is almost certainly caused by a partial delithiation. The delithiation is driven by the decomposition of fluorine containing polymer, where lithium reacts with the polymer to form LiF. The unit cell volume indicates that up to 180°C no reaction between PVDF and cathode occurs, since the volume of not heat treated precursor is also 33.8671 A3. Only at about 200°C the reaction starts and at about 300°C a major reaction occurs. We conclude that a film of LiF will be present at temperatures above 200°C. Base : less soluble base at higher treatment temperature. An optimum (lowest base) is observed at approx. 275-325°C. Soluble base is located on the surface, and dissolves into water to form LiOH or Li2CO3. Soluble base is the most reactive form of lithium. Thus, the lithium in the LiF which is formed by the reaction of fluorine containing polymer with the surface, will originate from the soluble base. In effect a LiF film replaces a film of soluble base. We observe that at least 250°C is needed to reduce the soluble base by 50%. At higher temperature ( > 325°C) new soluble base can re-form from the bulk, replacing the base which has been consumed by the LiF film formation. Moisture: a very low moisture content, together with a good moisture stability at > 200 to about 325°C. At temperatures above 325°C the polymer is gradually fully decomposing, and the surface is no longer protected against moisture uptake. At temperatures below 200°C the polymer does not fully cover the surface. Only at a sufficient high, but not too high temperature the surface is covered by a partly decomposed polymer film which protects against moisture uptake. It is obvious that a good coverage (=good wetting properties) are related to the reaction of polymer and soluble base on the surface. Figure 2a shows (TOP:)that the reversible capacity (CI: cycle 1) of the coated powder decreases whilst (MIDDLE:) the irreversible capacity (Qirr= [Discharge-Charge]/Charge, in %) increases significantly at temperatures above 300°C. At the same time (BOTTOM:) the rate performance (2C versus 0.1C, in %) deteriorates. There are 2 reasons for this observation: 1) Li is lost from the cathode to form LiF. If the oxygen stoichiometry equilibrates, then the loss of Li results in Li deficient-Li1-xM1+xO2- 1wt% PVDF contains about 6000 ppm fluorine, corresponding to a loss of about 3 mol% lithium. Generally, lithium deficient- Li1-xM1-xO2 has low rate performance and a high irreversible capacity; 2) The surface is covered by an electronically and ionically insulating LiF film, which is thicker than desired, which causes a poor rate performance. Figure 2b shows the results for the energy fade (capacity x average discharge voltage, measured at either 0.1C (TOP) or 1C (BOTTOM)) after cycling for 23 cycles between 3.0 and 4.5. Fig. 2b indicates an increase of cycling stability with increasing temperature until 250°C. The possibly improved cycling stability is almost certainly to be attributed to the formation of a protecting LiF film. Figures 3-5 shows the micrograph of the sample prepared at 200°C (SEM - Fig. 3), 250°C (FESEM - Fig. 4) and 350°C (FESEM - Fig. 5). Figure 3 shows the SEM of a sample prepared at 200°C: a particle is shown with many small "droplets" on the surface. The droplets are possibly molten PVDF particles. Evidently, the PVDF does not wet the surface. At 250°C (see Fig. 4) the drops disappear and the surface is smoothly covered by a PVDF film and surface structures indicate the formation of LiF plates below the film. At 350°C (Fig. 5) the polymer has fully decomposed and the surface is covered by small crystallic plates of lithium fluoride. Conclusion: Example 1 demonstrates that at a temperature above 200°C but below 350°C a polymer film covers the particles, where the interface between the polymer and the cathode surface is a film of LiF. The LiF film has replaced the soluble surface base of the cathode. Example 2: Example 1 investigated a coating by \% PVDF. However, at treatment temperatures T > 2758C, and especially > 300oC, it is observed that the decomposing polymer extracts so much Li from the cathode, causing a decrease of the reversible capacity. This indicates that the resulting LiF film might become unnecessarily thick. Therefore the present example illustrates the invention for a heat treatment using less polymer, only 0.3 wt% PVDF. As before, the example investigates the influence of temperature on the preparation of samples coated by polymer having an LiF interface. An LNMO mass production sample is used as cathode precursor. Its composition is Li1+xM1-xO2 with M=Ni0.5Mn0.3CO0.2 and x about 0.00. The precursor further contains 0.145 mol% 5 and 142 ppm Ca. 100g of cathode precursor and 3g of PVDF powder are carefully premixed using a coffee grinder. Then the 103g of mixture is mixed with the remaining 900g cathode precursor, and mixed at medium energy using a Haensel type mixer. The mixture is sampled to batches of 100g each. These batches are heat treated for 5h at temperatures ranging from 225 - 350°C. Samples are prepared at 225, 250, 275, 300, 325 and 350°C against a "blank' sample without PVDF. The resulting powders are sieved, and analyzed in a similar manner as in Example 1. Table 2 gives a summary of samples, preparation and results: Figure 6 shows (BOTTOM:) results for pH titration before (*) and after (?)f as well as (TOP:) moisture content after humidity exposure (5days, 30°C, 50%). It shows that, similar to Example 1, the PVDF treatment lowers the soluble base significantly at temperatures above 250°C and it protects (but with a lower effect) against moisture uptake, with an optimum at 250-275°C . Good coin cell test results are obtained over the whole temperature region. As the base content decreases a LiF layer forms, and it is assumed that this LiF is beneficial for improving safety performance and high voltage stability in full cells. It can be concluded that, compared to Example 1, the optimum base content is observed at 275-35CoC, the moisture content is lowest in a limited range around 250oC, and the electrochemical test results are excellent in the whole temperature range. Even if some effects can already be observed using 0.1 wt% of PVDF, it seems that 0.3 wt% PVDF is near to the lower limit to achieve the desired results, where 1 wt% PVDF could be the upper limit; combined with a heating temperature of 200 to 300° C. This analysis is further explored in Example 5a-d below. An optimum equilibrium between the desired effects on base content and moisture uptake, without negatively affecting the electrochemical results, is to be found between 0.5 and 0.8 wt% PVDF, independently of the tested lithium transition metal oxide composition. Example 3: This example investigates the influence of temperature on the preparation of L1COO2 samples coated by polymer having a LiF interface. The example discusses the voltage profile and micrestructure of a suitable LiCo02 to give further evidence for the conclusions of examples 1-2. The key conclusions are similar to examples 1-2: between 200 - 350°C a LiF film forms. The thickness increases with temperature. Otherwise, a LiF film cannot be retained at higher temperature. The example shows results for samples prepared by adding 1% PVDF polymer. A lithium cobalt oxide mass production sample is used as cathode precursor. Its composition is 1 mol% Mg doped LiCo02, having a mean particles size of 17 urn. 1000g of this precursor powder and 10g of PVDF powder are carefully mixed using a Hensel type mixer. The mixture is sampled to batches of 150g each. These batches are heat treated for 9h at temperatures ranging from 150 - 600°C. The resulting powder is sieved. The powders are analyzed by coin cell testing, SEM and conductivity. The SEM analysis shows an irregular coating of polymer at 150°C, becoming increasingly smooth and homogeneous as the temperature increases to 250°C. At 300°C the surface layer starts to change, and at 350°C a surface film is observed that seems to have inorganic characteristics, instead of a being a polymer coating. At 600°C the surface film is damaged and well formed crystals, possibly being LiF, are created. The creation of the crystals proves that LiF does not wet the surface at higher temperature. It is seemingly impossible to achieve a LiF film by direct high temperature synthesis. Figure 7a shows the SEM graphs of the sample at 300°C, Fig. 7b at 600°C. Note the presence of well-formed LiF crystals. Table 3 gives a summary of the electrochemical testing measurements. Table 3: Charge (QC), Discharge (QD) and Irreversible capacities (Q irr) of samples treated at different temperatures. Figure 8 shows the discharge voltage profile (4.3-3.OV, 0.1C rate) of the samples in Table 3 prepared with 1% PVDF at different temperature. Samples prepared at lower temperature (150°C, 200°C) show exactly the same discharge voltage profile. The profile is similar but has slightly lower capacity (about 1% less) than the reference (data not shown) which is the untreated sample used as precursor. The capacity values refer to the actual mass of the sample (thus it includes the weight of the polymer coating). The low T samples (150, 200°C) contain 1% PVDF coating layer, this explains the 1% lower capacity. The voltage profile is typical for LiCo02 with high Li:Co ratio, because no phase transition at 4.1V is detected. The 250°C sample shows a different voltage profile, typical for a LiCo02 having a poor rate performance. The polarization is larger (voltage depression) and the end of discharge is much less square type (more rounded). This is attributed to a LiF interfacial layer formed between the polymer coating and the LiCo02 surface. This LiF layer is fully covering the surface and has low ionic and electronic conductivity, causing the low rate voltage profile. With increasing temperature (300°C, 350°C) the capacity deteriorates dramatically. This clearly indicates the formation of a resistive LiF layer with increasing thickness which obviously covers the whole surface. However, if the preparation temperature is increased further, at 600°C we observe almost near full capacity, improved rate performance (not shown) and a clear phase transition at 4.1V. (Normally the 4,1V phase transition is only observed for Li deficient or stoichiometric LiCoO2). These data at 600° C show that a resistive LiF surface layer is absent. Obviously, at elevated temperature the homogeneous LiF surface layer is destroyed and large fractions of the surface is not covered by a LiF layer anymore. The data are fully consistent with the SEM which shows a damaged surface and the creation of larger LiF crystals. Example 3 demonstrates that at temperatures above the melting point of PVDF (140-170°C) a homogeneous polymer surface film forms. However, the temperature needs to be increased to more than 200'C and preferably 250°C before the reaction between polymer and cathode surface - creating the desired interfacial LiF film forms. However, if the temperature is too high, the protection layer is not active anymore. The LiF surface film then detaches from the surface and LiF crystals form. Example 3 also shows that the results achieved on a 1% Mg doped LiCo02 sample are comparable to the LNMO sample of Example 1. Example 4: Example 4 investigates the influence of temperature on the preparation of LNO type samples coated by polymer having a LiF interface. The example shows results for samples prepared by adding 0.3 and 1% PVDF polymer. A LiNio.aCoo.15Alo.05O2 sample, with 0.15 mol% S and 500-1000 ppm C, has been prepared from an alumina containing mixed transition metal precursor and LiOH in flowing oxygen at 5 kg scale in a pilot plant. The PVDF treatment was done similar as in examples 1 and 2. Table 4 summarizes the samples, preparation and results. Table 4: Samples, preparation and obtained results for high nickel cathode materials The table shows that the PVDF treatment improves moisture stability and, at T =250°C the initial base content is lowered considerably. At 150°C, compared to no PVDF, no decrease of base is observed, but at higher T, due to the consumption of base to form LiF the base content decreases. For this LNO composition, the moisture content is at its lowest for a treatment at 350oC. Compared to the untreated sample the rate of base increase during humidity exposure slows down. Similar as in examples 1 - 3 the capacity and rate deteriorates at higher T if 1% PVDF is used, whereas using 0.3% PVDF allows achieving good electrochemical results over the whole temperature region. Example 5a-d This example reproduces the results of Examples 1 and 2 for larger scale samples. These samples are additionally tested in polymer type full cells. In all Examples mass production LNMO (M=Ni0.5Mn0.3CO0.2) with Li:M of approx. 1.0 is used as precursor. The precursor further contains 0.145 mol% S and 142 ppm Ca. Example 5a: 1 wt% PVDF at 250°C 200 g mass production LNMO and 18g PVDF powder are pre-mixed in 4 batches using a coffee grinder. The mixture is added to 1.6 kg of LNMO and mixing continues using a Hensel type mixer using a 2 L vessel. The mixture is heat treated at 250oC in a convection oven for 5h, followed by sieving. Example 5b: 1 wt% PVDF at 250eC (larger sample) 15 kg of cathode precursor powder and 150g of PVDF powder are carefully mixed using a pilot plant ribbon blender. The powder mixture is heated for 5 h at 250°C followed by grinding and sieving. Example 5c: 0.3% PVDF at 300oC Basically similar as the 1.8 kg sample of Example 5a with the exception that the heat treatment temperature is 300°C and less PVDF (5.4g) is used. Premixing was done with 2 batches of 50g sample with 2.7g PVDF. Example 5d: 0.3% PVDF at 350° C. Similar as example 5c with the exception that the heat treatment temperature was 350oC. Tests were performed in a similar manner as in example 1-3, additionally 800 mAh wound pouch type cells are assembled and tested (such type of cell is described in e.g. the prior art of US 7,585,589). Table 5 summarizes the results. QD: discharge capacity; Rate: in % vs 0.1C, Base: before and after humidity chamber exposure. The table allows for the following conclusions: 1) 1% @ 250°C sample: It has the best moisture stability. The base does not increase during humidity exposure and the moisture content after humidity exposure is very low. However, the LiF film is thin, and the base content is only reduced by approx. 30 %. 2) 0.3% @ 300T sample: Caused by the thinner polymer film the moisture stability is worse than that of 1%@250oC, otherwise, the total base is low, less than 50% of the reference. This indicates that the LiF is better developed and the decomposition of the polymer has consumed most of the base. We observe a slight decrease of unit cell volume, consistent with the extraction of some lithium from the bulk. 3) 0.3% @ 350°C sample: the moisture content is better than at 300oC. Table 6 summarizes the pouch cell testing results. A dramatic decrease of swelling after high temperature storage (4h, 90°C) is observed. The swelling is the ratio of cell thickness after 4h measured when the cell is still hot (90°C) compared to the thickness measured before the test (cold). Several further tests with differently treated samples were performed, but only the PVDF treated samples show a dramatically reduced swelling, much lower than the typical obtained figures of 40-50%. We furthermore observe that all PVDF treated cells pass the overcharge test which indicates improved safety performance. Overcharge is done at 700 mA until 5.5V is reached. Passing means that no fire or smoke event happens. Nailing test is done using a 2.5 mm diameter sharp nail at a speed of 6.4 mm per second. Passing means no smoke or fire. Example 6: Example 6 Is a so-called 'blank' example, and simulates a possible reaction that happens between the molten PVDF covering the surface of the particle and the LiOH type base present on the particles" surface. By using a Differential Scanning Calorimetry (DSC) method, this example shows that polymer reacts with lithium containing base at temperature of about 50°C above the melting point of Kynar. This reaction is necessary to create the desired inner LiF layer. A Kynar® 2801 sample from Arkema (received as fine powder and having a melting point - as reported by the producer - of 142° C) and a LiOH*H2O sample are each jet milled until their average size (D50) is below 2 um. Figure 9 gives a SEM picture of the Kynar® sample, showing that it is composed of agglomerated ball-shaped primary particles having an average particle size of between 0.2 and 0.5 um. The resulting fine Kynar powder, and fine particles of LiOH*H2O are then mixed in a 2 : 1 mass ratio. This corresponds to a molar ratio F:Li of fluorine in the Kynar to Li in the hydroxide of approx. 2.62. Hence, even if all Li reacts with polymer, there is still an excess of unreacted polymer. This mixture is heated to 150, 200 and 2508C. The mass loss is recorded and X-ray diffraction is measured for the heated blends. The blend, and the Kynar reference are investigated by DSC. The samples are inserted into stainless steal DSC cans which are hermetically sealed. The heat flow is measured during heating, using a temperature rate of 5K / min from room temperature to 350°C. Figure 10 shows the obtained DSC results (heat flow vs. temperature; top: blend of Kynar and LiOH*H2O; bottom: pure Kynar): the minimum heat flow (most endotherm) for Kynar is achieved at 142'C, which is identical to the melting point of 142°C. The curve obtained during heating of the blend is completely different. First, a sharp endothermic event is observed with minimum heat flow at 109.1°C. This is the release of moisture LiOH*H2O ? LiOH + H2O. Then a strong exothermic event is observed. The maximum heat flow is observed at 186.2°C. It is assumed that at this temperature PVDF in contact with Li base and high pressure moisture decomposes and LiF (and possibly carbon) is formed. The DSC cans are hermetically sealed so no further reaction takes place. In air however at higher temperature the polymer will continue to decompose, as will be shown in Example 7. Example 7 This example is another 'blank' example, and simulates a possible reaction that happens between the molten PVDF covering the surface of the particle and the LiOH type base present on the particles' surface. The example shows that in air at temperatures above 200°C a reaction between base and PVDF happens, which causes the creation of decomposed polymer and possibly carbon. The sample blend as in example 6 (Kynar® 2801 sample from Arkema & jetmilled of LiOH*H2O in a 2:1 mass ratio) is used. The blend is heated to 150, 200 and 250oC in air for 5h. The mass loss is recorded and X-ray diffraction is measured for the heated blends. Please note that example 6 is a closed system (high pressure moisture) whereas example 7 is for an open system (where possibly most moisture evaporates). Table 7 summarizes the results, where 'X-ray' lists the observed compounds. At 150'C and 175oC the blend has basically not reacted. The color is white - yellowish and the electric conductivity is zero. A mass loss of 13-15wt% is observed, mostly originating from the the reaction LiOH*H2O + PVDF ?LiOH + PVDF. The blend is fully soluble in NMP. The X-ray diffraction pattern shows LiOH and LiOH*H2O, Li2CO3, polymer and traces of LiF. At 200°C the blend reacts. The resulting color is black. A much larger mass loss is observed. The conductivity could not be measured (too low). The blend cannot be fully dissolved in acetone, and black particles remain. The X-ray diffraction pattern shows LiF and polymer. The polymer has a different diffraction pattern than pure PVDF. At 250°C a stronger reaction occurs. The mass loss is 58.7 wt%. The blend shows an increased conductivity of 3*10-7 S/cm. The blend cannot be fully dissolved in acetone, and black particles remain. The X-ray diffraction pattern shows LiF and polymer, the polymer having a different diffraction pattern than pure PVDF. Figure 11 and 12 show some of the collected X-ray diffraction patterns: Fig. 11 shows a reference Kynar (PVDF) sample (below) and a sample treated at 250°C (top), Fig. 12 at 150 (lower curve of top figure) and 175°C (upper line of top figure), and again the reference PVDF on the lower figure. In Fig. 11 the two high intensity peaks of the top figure (only 10% of full scale shown) are LiF. The broad hump at 15-30 deg, is 'undefined' polymer, remaining from PVDF but having a clearly different X-ray pattern than the reference. The pattern of a blend after a heat treatment at 200"C (not shown) is very similar. In Fig. 12 the X-ray diffraction pattern of PVDF precursor (^reference) and of the blend after 150 and 175°C heat treatment show that basically PVDF does not react, and the pattern of PVDF remains. At 150°C tiny traces of LiF can be detected. At 175°C LiF becomes a minor impurity. Diffraction peaks of LiF are marked by an arrow. Other peaks can be indexed to lithium salts such as Li2CO3 and LiOH. Example 7 shows that in air, at about 200°C, i.e. about 50 Khigher temperature than the melting point, a reaction between LiOH and PVDF takes place which creates LiF and a modified polymer. The example confirms the model that the decomposition of PVDF and the formation of the LiF layer on a lithium transition metal oxide powder should be caused by the reaction of PVDF with lithium base. Example 8: This experiment is designed to prove the following: 1) at low heating temperature no LiF layer is present (PVDF just covers the particle but no LiF reaction layer is formed) 2) at the heating temperature according to the invention a reaction between PVDF and cathode is initiated (resulting in a thin interfacial LiF layer) 3) at too high temperature a thick LiF film has formed (all PVDF has been consumed by reacting with the cathode to form LiF). LiF has a small solubility in water (about 1.5g or so per L). On the other hand, PVDF is insoluble. Thus it is expected that after immersing heat treated product, the LiF dissolves, and any dissolved fluorine ions can be detected by liquid chromatography. However, as PVDF containing samples are hydrophobic, it is however not sure that all LiF will be accessible by water. Since PVDF is highly soluble in acetone or NMP but LiF is not, samples can also be prepared where the PVDF is removed by dissolution in NMP or acetone, to ensure that water can access and dissolve the LiF. The following samples were tested: 1) A sample as prepared - without washing - the same or similar samples as described or analyzed in Example 1, 2) A sample washed in a small amount of acetone and decanted , 3) A sample washed in NMP and decanted. The liquid chromatography (LC) procedure is as follows: 1) weigh 1g of sample in a glass Erlenmeyer flask of 300 mL; 2) add 100 mL of doubly de-ionized water; 3) add a glass stirring bar and stir for 1 hour; 4) filtrate over microfilter Millipore 0.45pm; 5) measure filtrate on ion-chromatograph (along with procedure blanks). Table 8 summarizes samples, preparation and results Fraction %: the % of PVDF that has reacted, as can be deduced from the amount of F found. The F(-) analysis results indicate 3 different levels in the wash water: 0.006 to 0.010 wt%, indicating that nearly no LiF is present; 0.054 to 0.064 wt%, indicating the presence of a LiF layer of nearly the same thickness; and finally 0.158 to 0.162 wt%, indicating nearly all of the PVDF has reacted, as will be explained below. The results are also shown in Figure 13. The upper panel shows the wt% of fluorine detected by chromatography. The lower panel shows the fraction of dissolved F detected by chromatography calculated from the above data. First we observe that washed (in NMP or acetone) and unwashed samples give the same result. Compare for example EX0121 and 0121C. Whereas PVDF offers an efficient protection to moisture uptake in the humidity exposure test, immersion in water enables underlying LiF to be dissolved. Secondly, the ionic chromatography clearly proves that at 150°C practically no (and in any case insufficient) LiF is present. See for example EX0121 and 0121C. Thus the polymer has not reacted with the surface of the treated cathode product. A PVDF film may cover the particle but the protecting LiF film does not exist. At 250°C a fraction of the PVDF has reacted, for example for EX0194 & 0194C. The total amount of LiF formed (= amount of reacted PVDF) does nearly not depend on the initial amount of PVDF, as is deduced from comparing EX0194C and 0126C. We conclude that the reaction rate is limited by the surface area of the cathode and by the availability of surface base. A large excess of unreacted PVDF covers the particle, but an interfacial layer of LiF has formed. At 350°C all PVDF has reacted. In an ideal experiment we would detect as much fluorine by LC as has been added to the sample in the form of PVDF. It is assumed that the obtained result for the fraction of detected fluorine (84-88%) is within the experimental systematic error and thus we conclude that at 350°C all PVDF has decomposed. It can be said that for the treatment at 250*C, the amount of F detected is surface limited, i.e. dependent on the quantity of base Li, whilst for 350°C the amount is possibly PVDF limited, i.e. dependent on the initial amount of PVDF. Example 9: This example describes the investigation of PVDF-treated cathode material using X-ray Photoelectron Spectroscopy (XPS) to investigate the decomposition of PVDF and the formation of LiF as a function of temperature. The example shows results for selected samples (EX0124, EX0127, EX0161) of Example 1 prepared by adding 1% PVDF and treatment at 3 different temperatures: 200°C, 250°C and 350°C. The experiment is designed to prove that: 1) Full decomposition of the PVDF coating is obtained by prolonged heating at high temperature (-350 eC). 2) With increasing temperature an increasingly thick LiF layer is formed. The fluorine in this layer is coming from the PVDF and the Li in this layer is coming from the surface base present on the cathode particle surface. The results of the C, F and Li spectra are summarized in Table 9. Table 9: Overview of apparent atomic concentrations (at%) measured at the surface after deconvolution of the C 1s, F 1s and Li 1s spectra into their different contributions. Conclusions for Table 9: 1 C1s: 1.1 Disappearance (^decomposition) of PVDF at 350° C shown by decrease in CF2-CF2- peak at 291.1 eV. PVDF (pristine or partly decomposed) remains present at temperatures below this temperature. 1.2 Li2CO3 observed at the particle surface by CO3-peak at 289.2 eV. At 350T, Li2CO3 is removed. This can be explained by the formation of LiF in which the Li2CO3 present at the surface of the particles is used as the source of Li. 2 F1s: 2.1 Disappearance (=decomposition) of PVDF at 350°C shown by decrease in F-org-peak at 686.7 eV. PVDF (pristine or partly decomposed) remains present at temperatures below this temperature. 2.2 Formation of LiF at 35CTC shown by LiF-peak at -685 eV. 2.3 The formation of LiF is directly linked to the decrease in Li2CO3 indicating the use of Li2CO3 during this formation. The formation of LiF at lower temperatures cannot be concluded due to masking of this LiF layer by the PVDF overlayer (knowing that XPS has a limited penetration depth). Therefore, in Example 8, the PVDF overlayer is removed by solvent wash. 3 Li 1s: 3.1 Decrease of LiVLiF ratio closer to 1 when temperature is increased and more LiF is formed. This clearly shows that at 350"C the formation of LiF is complete and all Li at the surface is present as LiF. The XPS data clearly support the model that 1 Full decomposition of the PVDF coating is obtained by prolonged heating at high temperature (-350 °C). 2 With increasing temperature an increasingly thick LiF layer is formed. The F in this layer is coming from the PVDF and the Li in this layer is coming from the surface base present on the cathode particle surface. (The surface base consists of lithium salts Like Li2CO3 and LiOH. The Li2CO3 is a major part of the surface base and can be monitored by XPS) specifically: 2.1 at low T (150-200°C) PVDF is still present as a coating and there is almost no LiF present. All surface base (LizCO3) is still present on the surface of the cathode material. 2.2 at elevated T (250"C) a reaction between PVDF and the Li2CO3 has started {resulting in a thin interfacial LiF layer). PVDF is also still present as a coating. 2.3 at high T (3508C) a thick LiF film has formed: over time the PVDF fully decomposes and its F is consumed by reacting with the available Li2CO3 at the particle surface, to form UF. Example 10: This example investigates a PVDF-coated cathode material using X-ray Photoelectron Spectroscopy (XPS) to investigate the decomposition of PVDF and the formation of UF as a function of temperature. This example gives results for 0.3% PVDF and treatment at 3 different temperatures, 150°C, 250°C and 350°C. Selected samples of Example 2: EX0120, EX0126 and EX0160 are investigated. XPS is a surface sensitive technique with limited penetration depth. In Example 9 the evolving underlying LiF interface was masked by the polymer surface and could be detected only for the high T sample where the polymer has decomposed. In the present example a washing step is applied to remove remaining PVDF and more clearly visualize the underlying LiF layer. Samples EX0120, EX0126 and EX0160 are washed using the following procedure: 1) 5g in 20 ml NMP, shaking 1h; 2) diluting with 40 ml acetone; 3) decanting 2 times, drying. Since polymer is soluble in NMP and acetone, but LiF has practically no solubility, we assume that polymer is removed and the underlying LiF is accessible for XPS analysis. The results of the C, F and Li spectra are summarized in Table 10. Figure 14a shows the F1s spectrum at 150°C, 14b at 250T and 14c at 350'C. Counts per seconds (CPS) is plotted against Binding Energy (eV). Table 10: Overview of apparent atomic concentrations (at%) measured at the surface after deconvolution of the C 1s, F 1s and Li 1s spectra into their different contributions. 7" stands for absence of an XPS peak. Conclusions for Table 10: 1 C 1s: 1.1 Based on the absence of CF2-CF2 peaks we can conclude that most of the PVDF is removed by the solvent wash. Especially at T - 350"C no PVDF is observed (due to complete decomposition and conversion into LiF). 1.2 Li2CO3 observed at the particle surface by CO3-peak at 289.7 eV. Direct link between removal of Li2CO3 and increase in temperature is explained by PVDF that is converted into LiF. In this process, the Li2CO3 present at the surface of the particles is used as the source of Li. 2 F 1s: 2.1 The increasing formation of the LiF layer with increasing temperature is clearly shown by the increase of the typical LiF peak at 684.7 eV (see Figure 13). 2.2 The formation of LiF is directly linked to the decrease in Li2CO3 indicating the use of Li2CO3 during this formation. 3 Li 1s: 3.1 Decrease of Li+/LiF ratio closer to 1 when temperature is increased and more LiF is formed. This clearly shows that at 350*C the formation of LiF is complete and all Li at the surface is present as LiF. At 250°C there are still some small amounts of other Li-species present such as Li2CO3. At 150"C there are mainly the other Li-species present and almost no UF. LiF thickness: LiF thickness calculations are based upon standard exponential attenuation of the photoelectron intensity as a function of traveled distance as described by van der Marel et al. in Journal of Vacuum Science and Technologies A, 23 (5) 1456-1470 (2005). It is assumed that the layer structure of the present samples is as follows: bulk MnOx, CoOx, NiOx, C in -CO3 and Li+resl / Li and F in LiF / organic C, organic F and O-org and that the LiF forms a homogeneous layer. LiF thickness increases as a function of temperature: at 150qC, an initial thin layer of only 0.2 nm has been formed. At 250*C, the LiF thickness has almost reached its full thickness, being 1 nm or more. At 350°C, the LiF layer has reached its full thickness and PVDF has been fully consumed. These results were comparable to thicknesses obtained from Fluor ion chromatography. Example 10 gives strong evidence that at sufficient high temperature - about 50°C above the melting point - the polymer starts reacting with the surface base and a protective LiF film is formed by consuming and replacing the surface base. Based on the results of Examples 1 to 10 it can be concluded that an effective LiF film should have a thickness of at least 0.5 nm (extrapolated value at >200eC), and preferably 0.8 nm (extrapolated value at >225°C). The invention can alternatively be described by the following clauses: 1. A lithium transition metal oxide powder for use in a rechargeable battery, wherein the surface of the primary particles of the powder is coated with a first inner and a second outer layer, the second outer layer comprising a fluorine-containing polymer, and the first inner layer consisting of a reaction product of the fluorine-containing polymer and the primary particle surface. 2. The lithium transition metal oxide powder of clause 1, wherein the reaction product is LiF, wherein the lithium originates from the primary particles' surface. 3. The lithium transition metal oxide powder of clause 2, wherein the fluorine in the reaction product LiF originates from partially decomposed fluorine-containing polymer present in the outer layer. 4. The Lithium transition metal oxide powder of any one of clauses 1 to 3, wherein the fluorine- containing polymer is selected from the group consisting of PVDF, PVDF-HFP, and PTFE. 5. The lithium transition metal oxide powder of any one of clauses 1 to 4, wherein the fluorine- containing polymer is composed of agglomerated primary particles having an average particle size of between about 0.2 and about 0.5 urn. 6. The lithium transition metal oxide powder of any one of clauses 1 to 5, wherein the lithium transition metal oxide is selected from the group consisting of: - LiCOaMe02, wherein M is either one or both of Mg and Ti, with e<0.02 and d+e=1; - Lii+aM'i-aOz±b M1k Sm with -0.030, b">0, c">0 and a"+b"+c"=1; and a'Vb" > 1. 8. The lithium transition metal oxide powder of clause 7, wherein 0.50, b">0, c">0 and a"+b"+c"=1; and a'Vb" > 1. 20. The process according to clause 19, wherein 0.50, b">0, c">0 and a"+b"+c"=1; and a"/b" > 1. 7. A lithium transition metal oxide powder according to claim 6, characterized in that 0.5=a"=0.7, 0.10, b">0, c">0 and a"+b"+c"=1; and a"/b" > 1. 15. A process according to claim 14, characterized in that 0.5=a"=0.7, 0.1