Mechanism of Gases Generation during Lithium-Ion Batteries Cycling (2023)

Electrochemical cell

Electrodes and electrolyte

On-line electrochemical mass spectrometry (OEMS)

Cells cycling

The cell cycling was fulfilled on the following scheme. The cells charge was executed in the mode CC/CV (constant current/constant voltage) in the voltages range 2.6-4.2V (first group of experiments) and in the voltages range 2.6-4.8V (second and third groups of experiments). Charging was performed by the current 0.2C with due account of the following theoretical capacities of the NMC electrodes: 150mAh/g_NMC for the 4.2V cell and 190mAh/g_NMC for the 4.8V cell.¹⁴ The step (application of the constant voltage (CV) at the upper cutoff potential) was performed until reaching the limiting current 0.05C. Discharging was performed in the mode CC (constant current 0.2C was applied). It should be observed that the graphite electrodes capacity is much higher than that of the NMC electrodes (theoretical capacity of graphite electrodes makes 370mAh/g_SLP30 (Ref.20)).

In the first group of experiments, the evolution of gases when cells were charged in the standard mode was studied. For this purpose, four charge/discharge cycles of the experimental cell (Fig.1) were performed in the voltages range 2.6−4.2V at the temperature 25°C. The cycling results are represented in the Fig.2.

From the Fig.2, it is seen that most intensively, the gases C₂H₄, CO, H₂ are evolved during the first hour of charging. This is connected with sharp growth of the cell voltage from 2.6V to 3.6V.

In the interval from the beginning of the charge first cycle to 4.0V, a sharp growth of ethylene concentration is observed approximately up to 10μmol m⁻²_SLP30, which corresponds quite well to the data of other experimental researches.^34–36 The ethylene concentration growth is a consequence of EC reduction during the process of the SEI layer formation on the graphite electrode:^20,35,37,38

and^33,39,40

or a consequence of SEI decomposition⁴¹

Upon reaching the voltage 4.0V, the sharp growth of the C₂H₄ concentration stops, which is usually connected (in researchers' opinion) with the SEI layer formation end.^20,21 However, as it is seen from the Fig.2, slowly, the C₂H₄ concentration grows even farther at the cell cycling. Notably it is going on independently on a cycling voltage, i.e. the ethylene release is potential-independent. Usually the ethylene concentration growth during the cell cycling is subjected to no analysis at all as it is thought that this concentration growth lies in limits of an experimental error.^20,21 Nevertheless from the obtained experimental data (Fig.2), for C₂H₄ concentration growth speed, we have the value 0.075μmol m⁻²_SLP30h⁻¹. This process will be discussed by us below in this section.

Now let us consider hydrogen evolution on the Fig.2. The consideration is widely accepted^{20–25,34,44} that the hydrogen evolution in a first cycle of charging till the voltage 4.0V reaching is connected with a residual moisture removal from the electrolyte. Indeed according to our experimental data (Fig.2) in the first cycle, while the voltage grows up to 4.0V, the hydrogen is evolved in amount of approximately 2.4μmol m⁻²_SLP30. With due consideration of the fact that at full decomposition of one water molecule (on any mechanism), one molecule of hydrogen is formed, i.e.

so for the water concentration in the electrolyte, we obtain the value 20.1ppm. This value (in limits of the experimental error) matches well to direct water concentration measurements in our electrolyte (see the section Electrodes and electrolyte). In the calculations, the following data were used: anode surface area on BET equal to 0.084m,² electrolyte density 1.5g mL⁻¹,Ref.45 electrolyte amount in the cell 120μL.

Later on during cycling, the hydrogen concentration grows and moreover independently on a cell voltage, i.e. the hydrogen evolution is potential-independent. In the papers,^20,21 it was experimentally proved that this hydrogen evolution is connected with the electrolyte decomposition. This problem will be studied by us in detail below (see the section Mechanism of electrolyte decomposition and gases evolution at lithium-ion cells cycling).

Let us give consideration to carbon dioxide evolution shown in the Fig.2. In the beginning of the charge first cycle, we observe CO₂ evolution and then its consumption. This process will be discussed by us below in this section.

An essential CO₂ evolution starts at cell charge voltage higher than 4.0V. In the papers,^{20,42,43,46–48} it was proved that at voltages higher than 4.0V on the cathode, there runs oxidation of Li₂CO₃with CO₂ evolution. The authors of those papers note that the oxidation of Li₂CO₃ runs according to the following electrochemical reaction

However with use of the reaction (Eq. 5), the authors did not come to a quantitative correspondence with the experimental data on the CO₂ release. Besides, both in their experiments⁴³ and in our ones (Fig.2), there was not observed an oxygen evolution at these voltages in compliance with the reaction (Eq. 5).

Below (see the section Mechanism of electrolyte decomposition and gases evolution at lithium-ion cells cycling) it will be shown that the evolving-on-cathode (because of Li₂CO₃ decomposition (Eq. 5)) very reactive atomic oxygen interacts in situ with CO evolving-on-cathode (because of electrolyte decomposition²⁰), which results in CO₂ formation, i.e.

Thus, with due account of the reactions (Eqs. 5 and 6), the total reaction of oxidation of Li₂CO₃ on a cathode will look like as follows

It should be mentioned that in the paper,²¹ it was experimentally proved that the adsorbed-on-cathode atomic oxygen is an active participant of generation of gases CO₂ and CO at electrolyte decomposition.

Based on the Equation7, let us evaluate an amount of the released CO₂. Earlier it was found that at the course of the cell cycling after reaching the voltage 4.0V (in the first cycle), the amount of C₂H₄grows slowly with the rate 0.075μmol m⁻²_SLP30h⁻¹. Hence, during a total cycle of cell charge/discharge (10h) in the electrolyte, will be dissolved Li₂CO₃ in amount of 0.75μmol m⁻²_SLP30 (Eqs. 1–3). For the same time, the total amount of Li₂CO₃ additionally dissolved in the electrolyte must get oxidized on the cathode according to the Equation7 and CO₂ formation must take place. Hence, for the total cycle, 1.50μmol m⁻²_SLP30 carbon dioxide must be evolved. This value in the limits of the experimental error matches very well to the CO₂ amount evolved within second, third and fourth cycles (Fig.2). Thus, the amount of CO₂ evolved on the second and next cycles has a strict quantitative explanation based on the electrochemical reaction (Eq. 7).

However in the first cycle, the amount of evolved CO₂ is much higher, approximately 2.4μmol m⁻²_SLP30 (Fig.2). As it has been shown earlier, on expense of the electrochemical reaction (Eq. 7) in each cycle, the carbon dioxide is formed in the amount of 1.5μmol m⁻²_SLP30. Hence, in the first cycle, the amount of additionally released CO₂ is 2.4-1.5 = 0.9μmol m⁻²_SLP30.

The first cycle differs from all the next cycles only with the fact that within this cycle, residual moisture is removed from the electrolyte. Therefore, on the cathode because of electrochemical decomposition of the residual moisture, the atomic oxygen must be adsorbed and additionally form CO₂ in accordance with the reaction (Eq. 6). For a formation of the deficient amount of the carbon dioxide (0.9μmol m⁻²_SLP30 based on the Equation6), the oxygen is needed only in the amount of 0.45μmol m⁻²_SLP30. Thus, only 0.45μmol m⁻²_SLP30 oxygen and not oxygen total amount (1.2μmol m⁻²_SLP30) obtained from residual moisture decomposition takes part in the formation of the deficient amount of CO₂. The remaining oxygen amount must participate in other chemical or electrochemical reactions.

An investigation in-detail of these chemical and electrochemical reactions likewise of all the processes running in the beginning of the first charge cycle is beyond the scope of this research. Nevertheless in order to give a complete picture, we would like to speak out our assumptions about the mechanism of these reactions.

In general at the present time, there is no full understanding of gases evolution processes running in lithium-ion cells in the beginning of the their first charge cycle. These processes were under investigation in the papers.^32,34 The authors put forward a hypothesis that in the cells charge beginning, the evolution of CO₂ runs through electrolyte decomposition on the anode (Fig.2). Then CO₂ consumption goes on expense of its reduction to the lithium oxalate.^46,47 However, the performed electrolyte study described in the paper⁴⁹ showed that cells cycling result in formation of neither lithium oxalate nor another product of electrolyte decomposition except for the following gases: CO, CO₂, H₂ and C₂H₄.

That is why taking our experimental studies experience as a guide, we incline to the opinion that more probable is the other mechanism of CO₂ evolution and consumption in the cell charge first cycle. In the beginning of the charge first cycle, a chemical reaction must take place between residual moisture and CO evolving on the anode²⁰ (because of electrolyte decomposition (Fig.2)), which results in CO₂ formation, i.e. in the water-gas shift reaction (WGS)Ref.50

The role of catalysts for the reaction WGS can be played by carbonates of alkali metals, in particular, Li₂CO₃ (Refs.51–54). The reaction runs in a basic environment, which also must be formed on the anode in presence of residual moisture⁵⁵

Thus, the reaction (Eq. 8) must run on the anode SEI layer during the cell charge first cycle as long as the electrolyte contains any residual moisture. For the common reaction WGS in virtue of the low solubility of CO in water, a high operating pressures are required for reaching of at least moderately active process on the alkali metals carbonates.^51–54 Meanwhile for the reaction (Eq. 8), such problem does not exist as CO is formed on the anode just in the place of reaction running as a result of the electrolyte decomposition.²⁰ Our preliminary experimental studies showed that in the first cycle beginning of the lithium-ion cell charge, indeed on the anode, the reactions (Eqs. 8 and 9) run. However, study in detail of these reactions is beyond the scope of this paper and will be a goal for our further investigations.

In sequel, with charge progress and lithium sufficient amount accumulation in the anode, the reaction must run of CO₂ reduction in the interaction with the intercalated lithium^56–58

The reaction (Eq. 10) will consume already evolved CO₂ (in compliance with the Fig.2) and – in addition – form the SEI layer on the anode.

Besides, the formed lithium hydroxide (Eq. 9) is a very effective CO₂ absorbent, in connection with what it is used in systems of breathing gas purification.⁵⁹

Thus, the reactions (Eqs. 8, 10, 11) explain evolution (Eq. 8) and consumption (Eqs. 10 and 11) CO₂ in the very beginning of the cell charge first cycle (Fig.2).

The proposed mechanism explains all the experimental data given in the Fig.2 on the CO₂ evolution in the beginning of the first cycle of the cell charge/discharge.

Undoubtedly, the proposed mechanism requires additional researches both experimental and theoretical. That is why in this paper, the proposed mechanism of gases evolution (Eqs. 8–11) in the beginning of the cell charge first cycle is to be considered as one of possible hypotheses.

Let us consider the carbon monoxide evolution on the Fig.2. In a quite a number of papers,^20,21,33 it was proved experimentally that CO evolution at a cell charge is connected with the electrolyte decomposition. This question will be studied by us in detail below (see the section Mechanism of electrolyte decomposition and gases evolution at lithium-ion cells cycling).

In the paper²¹ with use of ¹³C-labeled carbon, it was proved that CO and CO₂ evolution is not connected with an oxidation of the carbon additive into the cathode. That is why the observed CO and CO₂ evolution can be connected only with the electrolyte decomposition.

In conclusion, we would like highlighting the fact that in this group of experiments, the concentrations ratio is CO₂/CO = 0.82. This evaluation will be of importance for the discussion of the electrolyte decomposition mechanism.

In the second group of experiments, there was studied, how the upper cutoff voltage influences gases evolution velocity. For this purpose, four cell charge-discharge cycles were performed in the voltages range 2.6−4.8V at the temperature 25°C. The cycling results are represented in the Fig.3.

From the beginning of the first charge cycle and up to the voltage 4.0V (Fig.3), a sharp growth is observed of the ethylene concentration up to 10.1μmol m⁻²_SLP30. This value differs not much from results obtained in the first group of experiments and the results obtained in the other researches.^{20,21,34–36} This is nothing to be surprised about as the ethylene evolution is determined by the SEI layer formation and must not depend heavily on the upper cutoff voltage. In sequel at cell cycling, the C₂H₄ concentration grows – potential-independently – with the approximate speed 0.075μmol m⁻²_SLP30h⁻¹. Within the discussion of the first group of experiments, it was shown that this C₂H₄ concentration growth was connected with the dissolution of Li₂CO₃ from the SEI layer and further restoration of this layer with C₂H₄ evolution according to the reactions (Eqs. 1–3).

Then the dissolved Li₂CO₃ is oxidized on the cathode with parallel CO₂ evolution according to the reaction (Eq. 7). It should be noted that the C₂H₄ concentration growth speeds in the first and second groups of experiments are the same in limits of the experimental error. This is just what should be expected as the C₂H₄ concentration growth speed must depend only on Li₂CO₃ dissolution speed from the SEI layer, which is determined only by properties of Li₂CO₃ and the electrolyte at the same temperature. Hence, it must not depend on a value of the upper cutoff voltage, which indeed was shown by the experiment represented in the Fig.3.

Now let us give consideration to hydrogen evolution on the Fig.3. The hydrogen evolution from the beginning of the first charge cycle up to reaching the voltage around 4.0V is connected with the removal of the residual moisture from the electrolyte.^{20–25,34,44} According to our experimental data (Fig.3), from the beginning of the first charge cycle and up to reaching the voltage 4.0V, the hydrogen is released in amount of approximately 2.5μmol m⁻²_SLP30, which – in limits of experimental error – coincides with the similar value in the first group of experiments. Also this result should be expected as in both groups of experiments, the same electrolyte was used with the same residual moisture. But in this group of experiments as a result of the four charge/discharge cycles, the hydrogen was evolved in the amount of 26.2μmol m⁻²_SLP30, which is 2.5 times more than in the first group of experiments.

Let us give look to the CO₂ evolution in the Fig.3. From the Fig.3, it is seen that within the voltages range from 4.0V to 4.57V, the CO₂ concentration grows slowly and almost in linear fashion. However starting with the voltage 4.57V and up to the end of charging, the speed of the concentration growth sharply increases. This fact is an indication that the CO₂ evolution in these two voltages ranges is connected with two different electrochemical processes. Here and now we would like to attract readers' attention that the CO₂ evolution in this group of experiments at charge up to the voltage 4.57V is the same as in the first group of experiments. Indeed, within the voltages range from 4.0V to 4.57V, as it is known, the Li₂CO₃ oxidation runs on the cathode.^20,42,43,48 In view of the fact that in these experiments, the C₂H₄ concentration growth speed is the same as that in the first group of experiments, it follows that according to the reaction (Eq. 7), the same amount of CO₂will be evolved, i.e. approximately 1.50μmol m⁻²_SLP30. This value corresponds very well to the experimental data (Fig.3) at charge within the voltages range from 4.0V to 4.57V for second, third and fourth cycles of charge/discharge. In the first cycle of charge in the voltages range from 4.0V to 4.57V, the carbon dioxide is released in amount approximately 2.4μmol m⁻²_SLP30 just as in the first group of experiments. As it was shown earlier, the additional CO₂ evolution is connected with the residual moisture removal according to the reactions (Eqs. 6 and 8).

The coincidence of the evolved amounts of CO₂ in the first and second groups of experiments (within voltage range from 4.0V to 4.57V) follows immediately from the two facts given below. Firstly, in this voltages range, only one reaction takes place and this is the reaction of Li₂CO₃ oxidation on the cathode (Eq. 7) (Refs.42,43,48). The reaction (Eq. 7) speed is determined only by Li₂CO₃ dissolution speed in the electrolyte and it does not depend on the cutoff voltage at the same temperature. Secondly, the amount of residual moisture in the electrolyte in both experimental groups is the same. However it is exactly the amount of the residual moisture in the electrolyte, which determines the amount of additionally evolved CO₂ in the first cycle of charge according to the reactions (Eqs. 6 and 8).

The fundamental difference in CO₂ evolution on the Fig.2 and Fig.3 starts (at cell charge) from the voltage 4.57V and to the charge process end. From the Fig.3, it is seen that after reaching the voltage 4.57V, CO₂ evolution speed grows sharply. It is indication that from this voltage value, a new electrochemical process starts, a result of which is the CO₂ vigorous evolution. It should be observed that also after reaching the voltage 4.57V, the oxygen evolution starts (Fig.3). In the paper,²¹ the series of experimental proves were given, which establish the following facts. Firstly, after reaching the voltage 4.57V on the cathode surface, the highly reactive atomic lattice oxygen starts releasing from the cathode body. Secondly, the oxygen release is the reason of the electrolyte decomposition and the gases CO₂ and CO evolution.

Thus, the CO₂ evolution during cell charge in the voltages range, from 4.57V and up to the charging process end is connected with the electrolyte decomposition, an active participant of which is the atomic oxygen, releasing from cathode under these voltages. This process will be studied in-detail below in the frame of the discussion of the electrolyte decomposition mechanism (see the section Mechanism of electrolyte decomposition and gases evolution at lithium-ion cells cycling).

Let us give consideration to the O₂ evolution in the Fig.3. From the Fig.3, it is seen that the oxygen is released stepwise and besides, its concentration is decreased continuously. In our opinion, the mechanism of consumption of already released oxygen looks like as follows. After the atomic oxygen release from the cathode, two ways appear for the oxygen further reactions. Firstly, it can take part in the reaction of the electrolyte decomposition. Secondly, it can recombine and exit into the cell head space:

The oxygen obtained according to the reaction (Eq. 12) is measured with aid of OEMS inside of the cell head space (Fig.3). After charging, the oxygen release from the cathode stops. Then the oxygen located in the cell head space can return on the cathode (Fig.1) and be adsorbed on it (the reverse reaction (Eq. 12)). Then the adsorbed oxygen will take part in the chemical reaction (Eq. 6), which consumes the oxygen and decreases its concentration in the cell head space. Besides, in each charge cycle, the amount of the released oxygen (i.e. oxygen concentration top peaks on the Fig.3) also falls. According to the researches,²¹ this is connected with the oxygen exit from the cathode superficial layer and its structure change (from the layered structure to the spinel structure or rock-salt structure). Such structure contains much less oxygen. That is why in the subsequent cycles, it gets much harder for the oxygen to leave the cathode with increase of the oxygen-poor superficial layer.

In conclusion, we would like to highlight several features of the composition of the released gases mixture in this group of experiments. Firstly, the concentrations ratio is CO₂/CO = 2.4. Secondly, the amount of the released gases CO₂+CO in this group of experiments is 2.6 times more than in the first group of experiments. Hence, the intensity of the electrolyte decomposition sharply grows at increase of the upper cutoff voltage, above the voltage at which the atomic oxygen starts releasing from the cathode.

In the third group of experiments, the temperature impact was studied on the gases evolution. For the clarification of this question, four cell charge/discharge cycles were performed in the voltages range 2.6−4.8V at the temperature 60°C. The cycling results are represented in the Fig.4.

From the beginning of the first charge cycle and up to the voltage 3.65V, a sharp growth is observed of the ethylene concentration – up to 23.5μmol m⁻²_SLP30, which is 2.3 times more than it be in the second group of experiments during the formation of the SEI layer. Hence at the temperature 60°C, the formed SEI layer is 2.3 times thicker than at the temperature 25°C. Afterwards at further cell cycling, the concentration C₂H₄grows (potential-independently) with the speed 0.61μmol m⁻²_SLP30h⁻¹, which 8.2 times more than at the temperature 25°C. As it is known,^60–62 the standard electrolyte LP57 is not able to form the stable SEI layer at high temperatures without use of additives (such as LiBOB). This is exactly, what this experiment shows.

Now let us consider hydrogen release in the Fig.4. From the beginning of the first charge cycle and up to reaching the voltage 4.0V, a sharp growth of hydrogen concentration is observed – approximately up to 5.5μmol m⁻²_SLP30, which is 2.2 times more than in the second group of experiments. In all the experiments, the same electrolyte was used by us. Hence in this experiment already on the stage of the sharp hydrogen concentration growth, the processes take place of not only removal of residual moisture from the electrolyte but also the decomposition of the electrolyte itself. Then at cell cycling, the H₂ concentration grows potential-independent. Total in the four charge/discharge cycles, the hydrogen was released in the amount of 35.0μmol m⁻²_SLP30, which is 1.3 times more than in the second group of experiments. This fact is an indication of an intensification of the electrolyte decomposition processes with the temperature increase up to 60°C.

Let us give consideration to the CO₂ evolution in the Fig.4. The amount of the evolved CO₂ in the voltages range of the electrochemical reaction (Eq. 7) (from 4.0V to 4.57V (Fig.4)) is approximately the same that we see in the second group of experiments (Fig.3). The velocity of the electrochemical reaction (Eq. 7) is determined by Li₂CO₃solubility in the electrolyte. Hence the Li₂CO₃solubility in the electrolyte is not too much change in the temperatures range from 25°C to 60°C.

In the voltages range corresponding to the electrochemical reaction of electrolyte decomposition (running at voltages from 4.57V up to charging session end with aid of the released-from-cathode reactive atomic oxygen), the amount of the released gas CO₂ is approximately the same as in the experiments given in the Fig.3. Hence also the general amount of the released gas CO₂ in this group of experiments (29.1μmol m⁻²_SLP30 (рис. 4)) is approximately the same as in the second group of experiments (Fig.3). This is connected with the fact that in the second and third groups of experiments, the amount of the released-from-cathode oxygen is also approximately the same (Figs. 3 and 4). Below (see the section Mechanism of electrolyte decomposition and gases evolution at lithium-ion cells cycling) it will be shown that the reactive atomic oxygen is the reason of the CO₂ evolution. That is why the amount of the released gas CO₂ is determined in full by the amount of the released-from-cathode atomic oxygen.²¹ Hence the velocity of the atomic oxygen diffusion from the cathode does not change much at a cell temperature growth – from 25°C to 60°C.

In conclusion, we would like to highlight several features of gases evolution in this group of experiments. Firstly, the concentrations ratio is CO₂/CO = 1.2. Secondly, the amount of the released gas mixture CO₂+CO (53.9μmol m⁻²_SLP30) in this group of experiments exceeds 1.3 times the same in the second group of experiments. Hence the intensity of the electrolyte decomposition grows with the cell temperature increase.

An electrochemical decomposition of any electrolyte can take place only within a period of cell charge. Hence, if in the course of the electrolyte decomposition, gases are released, they are supposed to be released only over the period of the cell charge. However in our experiments (Figs. 2–4) all through the cycling, some gases (H₂, C₂H₄) were released during both cell charge and discharge, i.e. in the potential independent manner. Hence, in the experiments (Figs. 2–4) onto the classical process of the electrolyte decomposition, an additional process is overlapped, which changes the pattern of the gases evolution.

The purpose of this group of experiments is to study this unusual phenomenon.

This phenomenon was also experimentally established in the papers.^20,21,28,34 In the paper,²⁰ the authors take the view that the apparent potential-independence of the H₂ evolution is caused by the rate-limiting reduction of protic electrolyte decomposition species on anode. However the rate-limiting reduction of protic species on anode can result only in the fact that on the classical steplike curve of the gases evolution at cell cycling, the steps will become lower and less-pitched. This limitation cannot lead to the hydrogen evolution on the stage of a cell discharge as protic species are reduced on anode with the H₂ evolution only on the stage of cell charge. However the experiments given in the Figs. 2–4 show that hydrogen is released on stages of cell charge and discharge, i.e. completely potential-independently.

Prior to a discussion of the potential-independence of the gases H₂ and C₂H₄ evolution in the Figs. 2, 3, let us note that CO₂ is released only on cathode²⁰ during cell cycling. Notably, the CO₂ evolution on cathode has the classical stepwise pattern, while the gases H₂ and C₂H₄, which are released only on anode,²⁰ have potential-independent nature of their evolution. The carbon monoxide is released on both anode and cathode.²⁰ Along with it, its evolution has partially potential-independent and partially stepwise pattern (Figs. 2 and 3). From this fact, the conclusion can be drawn that all the gases released on cathode are featured with the stepwise pattern of their evolution, while for the gases released on anode the potential-independent pattern of their evolution is typical. Hence, the potential-independent pattern of gases evolution is connected not with properties of the gases evolution process (as the authors of the paper²⁰ think) but instead with anode properties.

An active mass of anode, on which gases are released in potential-independent manner, is represented by a finely dispersed powder of graphite. As it is known, the finely dispersed powder of graphite is able to adsorb gases in abundance.⁶³ That is why we take the view that the ability of the finely dispersed powder of graphite to absorb gases ensures the potential-independent pattern of their evolution. Indeed, at a cell charge, when electrolyte decomposition and gases evolution take place, a part of the gases released on anode will be adsorbed in pores of the graphite powder, while the other part of the gases will exit into the cell head space and be measured by OEMS. While a cell is discharging (electrolyte does not decompose and gases are not generated), the accumulated in graphite pores gases will exit into the cell head space due to the diffusion processes from the porous graphite. Hence in the case of gases adsorption on anode, their concentration must rise in the cell head space on both stages of charge and discharge, i.e. potential-independently.

In the paper,⁶⁴ it was shown that in the case of pure carbon-based materials, it is impossible to reach a high gravimetric capacity of hydrogen accumulation. An increase of the gravimetric capacity of hydrogen accumulation in carbon-based materials is possible only either by way of their structure change (in particular by way of reducing the pore size^65,66) or by way of doping of carbon-based materials.⁶⁷

In the case of graphite, a considerable growth of gravimetric capacity, the hydrogen accumulation takes place in the case of use of finely dispersed powders, which can be obtained by milling in ball mills.^68–70 Any imperfections of crystalline structure (particularly dislocations) are traps for hydrogen, as they decrease the energy of hydrogen atom as compared to their location in normal interstice. Besides they are the centers of hydrogen absorption, and also contribute to hydrogen penetration into the graphite depth. Hence, imperfections of the graphite crystalline structure cause sharp rise of hydrogen adsorption on it. It should be noted that for manufacturing of anodes of lithium-ion cells, exactly finely dispersed powders of graphite are used.

In quite a number of papers, it was shown that doping of carbon-based materials by alkaline or transitional metals results in a considerable growth of their gravimetric capacity of hydrogen accumulation.^67,68,71 This is connected with a general energy rise of hydrogen binding with the doped graphite powder.⁷¹ It should be noted that on anode, the graphite powder is doped by lithium. To the full extent, the said has relation also to other gases released on anode. However, the hydrogen cannot only adsorb on the graphite powder surface but also intercalate inside of the graphite and be accumulated there in abundance.^72–74 The ability of the hydrogen to be accumulated in electrodes is proved also by the studies,^24,25 in which it was shown that the hydrogen was released at lithium-ion batteries storage in their charged state.

If our assumptions are correct, at both cell discharge and charge before reaching of voltage value of electrolyte decomposition (4.57V), hydrogen is not supposed to be generated. Hence, it can get in the cell head space (and augment the H₂ concentration there) only due to its diffusion from the graphite powder pores. Hence, if (after cell charge completion) to interrupt the cycling and let hydrogen leave the graphite powder, then at subsequent discharge&charge to the voltage value 4.57V, the hydrogen concentration is not supposed to rise in the cell head space. With this purpose in mind, there were executed four cell charge/discharge cycles in the range of voltages 2.6-4.8V at the temperature 25°C. Notably, after second and third charge cycles, a pause was done for 10h (equal to the cycle duration). This pause is enough so that the hydrogen diffusion process from the graphite powder into the cell head space had time to be completed. The results of the cycling are represented in Fig.5.

The conducted experimental studies prove categorically that the potential-independence of evolution of the gases (H₂, C₂H₄ and CO) on anode is connected with the gases adsorption in the pores of anode graphite powder. It should be noted that in our previous paper⁷⁵ by direct experiments, it was proved that during lithium-ion cell cycling, the hydrogen is accumulated in anode graphite powder.

Before establishing the mechanism of the electrolyte decomposition and gases evolution at lithium-ion cells cycling, let us (based on the obtained-to-the-date experimental data) put in words criteria for which this mechanism must meet.

• 1. In the experiments conducted in the frame of the study,²⁰ the Li+-ion conductive glass ceramic was used, which was impermeable for gases generated in cathode and anode departments of a cell. This allowed – during cell cycling – analyzing generated gases separately in cathode and anode departments. As a result, it was proved that at cycling of lithium-ion cells and electrolyte decomposition, the following gases are released: on cathode CO₂ and CO, while on anode H₂ and CO. Besides, on anode, ethylene is released, too, but in the paper,²⁰ it was proved that the C₂H₄ evolution is connected with formation of SEI layer and has no relation to the electrolyte electrochemical decomposition.

• 2. Besides, the use of the Li+-ion conductive glass ceramic allowed proving²⁰ that in general, the electrolyte decomposition is determined by the interaction between the processes on cathode and anode, which often is referred to as "crosstalk".^76,77 It means that the electrolyte is oxidized on cathode and afterwards the oxygenated ions of the electrolyte move to anode and are reduced on it. Upon that, the gases evolution rises sharply on both anode and cathode as compared to cells, where the anode and cathode departments are separated by the Li+-ion conductive glass ceramic.

• 3. At electrolyte oxidation on cathode, also the acidity grows of the surrounding electrolyte.^20,78 In the paper,⁷⁸ it was shown experimentally that at the electrolyte oxidation on cathode, a high concentration of protons is generated on the cathode surface and the protons promote a corrosion of aluminum current collectors. Along with it the protons motion to anode together with the Li+ ions reduces the coulombic efficiency at cycling.^76,78

• 4. In the paper,²¹ a number of experimental proves was given that the release from cathode of highly reactive atomic oxygen is the main reason of the electrolyte decomposition and the CO₂ and CO evolution. Firstly, in all the experiments,²¹ the CO₂ and CO evolution occurs only after that the oxygen evolution can be observed. Secondly, on these or that cathodes (of different composition), the oxygen is released at less or higher voltage values than 4.57V; then at the same voltages, the electrolyte is decomposed and the gases CO₂ and CO are released.^21,32 More to it, if the atomic oxygen is released in a less amount, then equally less will be the evolution of CO₂+CO (Ref.21).

Here it should be noted that the chemical reaction between oxygen and ЕC (ethylene carbonate) at room temperature is possible only in the case that the oxygen is in its highly reactive form, for example, in the form of the atomic oxygen²¹ as ЕC does not decompose in dry air at working temperature of the lithium-ion cells.

From this, it follows that the electrolyte decomposition is a complicated electrochemical and chemical process, a very important stage of which is one of the chemical oxidation of the electrolyte by very reactive atomic oxygen.

Also in the paper,²¹ it was proved that due to the purely electrochemical process of the electrolyte decomposition (without the stage of the chemical oxidation of the electrolyte by the atomic oxygen), not more than 10% of the entire gas CO₂+CO obtained in the experiments can be generated.

• 5. In the paper⁴⁹ with aid of the method GC-MS (Gas chromatography–mass spectrometry), the electrolyte of lithium-ion cells was examined after a long cycling. It was found out that in the electrolyte, no new components occur connected with the electrolyte decomposition except for the gases H₂, CO₂, CO. Hence, as a result of cells cycling, the electrolyte decomposes in full onto the gases H₂, CO₂ and CO. To the same conclusion, we came, too. We obtained it with the method GC-MS, while studying various electrolytes used in lithium-ion cells after them long cycling.

• 6. The ratio CO₂/CO can vary (at the electrolyte decomposition as a result of cells cycling in the wide range from 0.82 to 2.4 (Figs. 2–4 and (Refs.20,21)) depending on temperature and the upper cutoff voltage.

Let us start our discussion of the electrolyte decomposition mechanism at cells cycling from consideration of the last criterion. Only two ways are possible, in which the ratio CO₂/CO would vary considerably in dependence with a temperature and an upper cutoff voltage.

Firstly, it is possible to make an assumption that two groups of concurrent electrochemical reactions exist. In the first group of the reactions, mainly CO₂ is generated, while in the second group, mainly CO is generated. Thus one can make an assumption that in dependence with a temperature and an upper cutoff voltage, one of the reactions groups receives an advantage and the ratio CO₂/CO changes much. However if to look at the formula of our electrolyte – C₃H₄O₃ (EC), – it is seen that it is impossible to write two different overall reactions for two electrochemical processes of the electrolyte (EC) decomposition so that as a result of one of them, mainly gases CO₂ and H₂would be generated, while as a result of the second one, mainly gases CO and H₂would be generated. It should be noted that at the electrolyte (EC) decomposition, no other components are supposed to be generated except for the gases CO, CO₂ and H₂ (the criterion 5). Thus, this way of the electrolyte decomposition is impossible.

Secondly, it is possible to make an assumption that there exists only one electrochemical reaction of the electrolyte decomposition and along with it, the CO₂ formation is a consequence of an additional chemical reaction running between the releasing from cathode of the reactive atomic oxygen (criterion 4) and the releasing – at the same place on cathode – of the gas CO (because of the electrolyte decomposition) (Eq. 6). In this case, it is unambiguously possible to write the overall reaction for the electrochemical process of the electrolyte (EC) decomposition meeting all the mentioned above criteria and this can be done as follows:

As it was highlighted earlier based on the studies specified in the criterion 4, the process of the electrolyte decomposition is a complicated multistep process running on cathode and maybe also on anode. Nevertheless based on the overall reaction (Eqs. 13) and on compliance with all the criteria, for sure, it is possible to write the total electrochemical reactions running separately on cathode and on anode.

The electrochemical reactions (Eqs. 14 and 15) (with due account of the reaction (Eq. 6)) meet the criteria 1–3,5,6. As a result of these electrochemical reactions, the electrolyte decomposes and only the gases H₂, CO and CO₂ are formed (with due account of the reaction (Eq. 6)) (criterion 5). On anode, the gases H₂ and CO are released, while on cathode the gases CO and CO₂ (with due account of the reaction (Eqs. 6)) (criterion 1). As a result of the electrolyte decomposition, its acidity rises (criterion 3). In general, the processes running on cathode and anode are interdependent (criterion 2).

However, as it was shown in the paper,²¹ a direct electrochemical decomposition of electrolyte (Eqs. 14 and 15) is unlikely and it is able to make only a small contribution into the general process of the electrolyte decomposition. Much more intensively, the process of the electrolyte decomposition runs, when prior to the electrochemical stage (Eqs. 14 and 15), the chemical process takes place of the electrolyte oxidation by the very reactive atomic oxygen released from cathode.²¹ In this case, the overall reaction (Eqs. 13) must be re-write in the form:

as the very reactive atomic oxygen released from cathode can be reduced only by the hydrogen coming from the electrolyte, which promotes the electrolyte (EC) decomposition. In this case, the electrochemical reactions (Eqs. 14 and 15) will accept the following form:

The resulted from the electrolyte decomposition water will be decomposed due to the additional electrochemical reaction (see below).

As because of the electrolyte decomposition its acidity must grow (criterion 3), the water is supposed to be decomposed on the acidic-type mechanism:

Upon that the H⁺ ions motion to anode together with the Li+ ions reduces the coulombic efficiency at cycling according to the experimental results represented in the paper.⁷⁶

From a comparison between the electrochemical reactions (Eqs. 17 and 18) and (Eqs. 14 and 15), it becomes evident that running of the electrochemical reactions (Eqs. 17 and 18) is more energetically profitable than running of the electrochemical reactions (Eqs. 14 and 15). Thus in the case of the atomic oxygen release from the cathode, in general, oxygen-type mechanism of the electrolyte decomposition must take place, i.e. the reactions (Eqs. 17–20). These reactions meet all the described above criteria 1–6. However in the case of atomic oxygen absence on cathode, the direct mechanism of the electrolyte decomposition is possible, too (criterion 4), i.e. the reactions (Eqs. 14 and 15). Also these reactions meet all the criteria 1–6. With the aid of the reactions (Eqs. 13–20), it is possible to explain all the experimental results obtained in this paper and other papers.^20,21

The main purpose of this section is to explain the results of the first group experiments (Fig.2) quantitatively based on the established mechanism of electrolyte decomposition (Eqs. 7, 13–20).

First of all, let us evaluate the ratio of the released gases (CO₂+CO)/H₂ based on the experimental data given in the Fig.2. According to both the direct mechanism of electrolyte decomposition (Eqs. 13–15) and the oxygen-type of the same (Eqs. 16–20), this ratio should be equal to 1.5. While estimating the ratio (CO₂+CO)/H₂, one should take into consideration only the gases released as a result of the electrolyte decomposition. Besides, as a part of the released gases are adsorbed on anode (Fig.5), an estimation of the ratio must be performed only when the adsorbed gases would leave the anode. Let us perform this estimation at the charge voltage 4.0V in the last cycle. According to the investigations described in Figure5, from this voltage level, the electrolyte decomposition starts and on anode the least possible amount of the adsorbed gases stays.

As for hydrogen, from the total amount of the hydrogen released before reaching the charge voltage 4.0V in the fourth cycle (9.0μmol m⁻²_SLP30. see Fig.2), it is needed to subtract the hydrogen amount released because of the decomposition of a residual moisture (2.4μmol m⁻²_SLP30, see Fig.2). Hence, only due to the electrolyte decomposition, there was released hydrogen in amount of 9.0-2.4 = 6.6μmol m⁻²_SLP30.

Now let us evaluate the amount of the gases CO₂+CO released only due to the electrolyte decomposition. For the reaction (Eq. 7), a half of CO₂ amount is generated due to the electrochemical decomposition of Li₂CO₃ (Eq. 5), while the second half of CO₂ amount is generated due to the electrolyte decomposition with CO subsequent oxidation on cathode under action of the atomic oxygen from the reaction (Eqs. 5), i.e. it is generated on the oxygen-type mechanism of the electrolyte decomposition (Eqs. 17–20). Hence, from the CO₂ total volume released within the cycle (1.5μmol m⁻²_SLP30, see Fig.2), only a half is released due to the electrolyte decomposition. That is why from the total amount of the gases CO₂+CO released before reaching the charge voltage 4.0V in the fourth cycle (13.0μmol m⁻²_SLP30, see Fig.2), it is needed to subtract the CO₂volume obtained due to the Li₂CO₃ decomposition in the three previous cycles, i.e. 1.5·3/2 = 2.25μmol m⁻²_SLP30. Besides, the additional carbon dioxide in the amount of 0.9μmol m⁻²_SLP30 is obtained because of the residual moisture removal (see the section Results of cells cycling in the voltage range 2.6-4.2V at 25°C). Also this carbon dioxide amount is to be subtracted from the total amount of CO₂. Hence, only the released due-to-the-electrolyte-decomposition amount of the gases CO₂+CO is equal to 13.0 -2.25-0.9 = 9.85μmol m⁻²_SLP30.

As a result for this ratio, we obtain the value (CO₂+CO)/H₂ = 1.492. This value coincides with the theoretical value 1.5 in the limits of relative experimental error (less than 1%). Thus the proposed electrolyte decomposition mechanism (Eqs. 7, 13–20) gives a very good correspondence between the experimental data and the theoretical calculations.

Now let us give consideration to the experimental ratio CO₂/CO≈0.82 for the released gases CO₂ and CO within all the four cycles. According to the data given in the Fig.2 during all four cycles of cell charge/discharge, the gas CO₂was released in amount of 6.9μmol m⁻²_SLP30, while CO of 8.4μmol m⁻²_SLP30. According to the studies, due to the electrochemical decomposition of Li₂CO₃ (Eq. 5) over the period of four cycles, the carbon dioxide was released in amount of 1.5·4/2 = 3μmol m⁻²_SLP30 (see the section Results of cells cycling in the voltage range 2.6-4.2V at 25°C). The same amount of CO₂ (3μmol m⁻²_SLP30) was released on the oxygen-type mechanism (Eqs. 17–20) due to the electrolyte decomposition with the subsequent CO oxidation (on cathode) by the atomic oxygen from the reaction (Eq. 5). Besides, it was shown that the additional 0.9μmol m⁻²_SLP30 carbon dioxide is released due to the residual moisture removal according to the reaction (Eq. 19) and possibly to the reaction (Eq. 8) (see the section Results of cells cycling in the voltage range 2.6-4.2V at 25°C). Therefore, in accordance with the calculations, the total amount of released gas CO₂should be 3+3+0.9 = 6.9μmol m⁻²_SLP30, which is fully consistent with the experimental data.

Simultaneously with the release of CO₂ on cathode, on anode on the oxygen-type mechanism, the CO amount must be released twice less (Eqs. 17–20) than that of CO₂ i.e. 1.5μmol m⁻²_SLP30. Hence, the remaining carbon monoxide in amount of 8.4-1.5 = 6.9μmol m⁻²_SLP30 is released due to the direct mechanism of electrochemical electrolyte decomposition (Eqs. 14 and 15). Indeed, at cells cycling until the upper cutoff voltage (4.2V at 25°C), the atomic oxygen from cathode is not released (Fig.2 and (Ref.21)). Hence, an electrolyte decomposition on the oxygen-type mechanism (Eqs. 17–20) is impossible.

Thus, in the first group of experiments on the oxygen-type mechanism (Eqs. 17–20), only a small amount of gases CO₂ (3μmol m⁻²_SLP30) and CO (1.5μmol m⁻²_SLP30) is released using oxygen from the reaction (Eqs. 5). The main amount of CO (6.9μmol m⁻²_SLP30) is released due to the direct mechanism of electrochemical electrolyte decomposition (Eqs. 14 and 15). That is why the ratio CO₂/CO≈0.82 (Fig.2) has a rather small value.

The main purpose of this section is to explain the results of the second group experiments (Fig.3) quantitatively based on the established mechanism of electrolyte decomposition (Eqs. 7, 13–20).

Now we'll give consideration to cells cycling up to the upper limit voltage of 4.8V at 25^оC (Fig.3). Prior to the charge voltage 4.57V reaching (in the first cycle), the gases evolution is the same as in the experiments of the first group (Fig.2). After the voltage value 4.57V is reached, the release starts of the atomic oxygen from cathode (Fig.3 andRef.21). Hence, after the voltage 4.57V, the electrolyte decomposition is supposed to run on the oxygen-type mechanism (Eqs. 17–20). As a result, the portion of CO₂ rises sharply in the total balance of the gases CO and CO₂; so the ratio CO₂/CO becomes equal to 2.4 (Fig.3).

Let us estimate several ratios for the gases released due to the electrolyte decomposition. First of all, let us evaluate the ratio of the released gases (CO₂+CO)/H₂ based on the experimental data given in the Fig.3. According to Fig.3 (prior to the charge voltage 4.0V reaching in the fourth cycle), the hydrogen was released in the amount of 22.58μmol m⁻²_SLP30, while CO+CO₂ in amount of 33.3μmol m⁻²_SLP30. It should be noted that the amount of hydrogen released due to residual moisture decomposition in this group of experiments is the same as in the first group of experiments, i.e. 2.4μmol m⁻²_SLP30. This is because the same electrolyte was used (see the section Results of cells cycling in voltages range 2.6-4.8V at 25°C). Hence, only due to the electrolyte decomposition, hydrogen was released in the amount of 22.58-2.4 = 20.18μmol m⁻²_SLP30.

Now let us estimate the amount of gases CO+CO₂ (prior to the charge voltage 4.0V reaching in the fourth cycle) released only due to the electrolyte decomposition. In this group of experiments, the amount of CO₂ released due to the electrochemical decomposition of Li₂CO₃ (Eq. 5) (in the first three cycles) is the same as in the first group of experiments (i.e. 1.5·3/2 = 2.25μmol m⁻²_SLP30) as it is determined by the solubility of Li₂CO₃ (which in the experiments of the first and second groups is the same) and it does not depend on an upper cutoff voltage (Fig.3). Besides, the amount of CO₂ released due to the residual moisture removal is also approximately the same (0.9μmol m⁻²_SLP30) because the same electrolyte was used (see the section Results of cells cycling in voltages range 2.6-4.8V at 25°C). Hence, only due to the electrolyte decomposition, the gases CO₂+CO were released in the amount of 33.3-2.25-0.9 = 30.15μmol m⁻²_SLP30.

As a result for this ratio, we obtain the value (CO₂+CO)/H₂ = 1.494which corresponds very well to the theoretical value 1.5 (Eqs. 16–20) (the relative experimental error is less than 1%). Hence, also in this group of experiments, the proposed mechanism of the electrolyte decomposition (Eqs. 7, 17–20) shows a very good correspondence between the experimental data and the theoretical calculations.

Second, let us estimate ratios for the gases released due to the electrolyte decomposition in the first and second groups of experiments. With due account of the obtained data, the ratio between the amount of CO+CO₂ released only due to the electrolyte decomposition in the second and first groups of experiments (prior to reaching the charge voltage 4.0V in the fourth cycle) will be equal to (CO+CO₂)₂/(CO+CO₂)₁ = 3.061, while the ratio for the amount of releasing hydrogen will be equal to (H₂)₂/(H₂)₁ = 3.058. From comparing these ratios, it is possible to draw a number of conclusions. Firstly, at growth of an upper cutoff voltage up to 4.8V, the rate of the electrolyte decomposition grew three times. Secondly, the amount of the gases CO+CO₂ and H₂grew by the same value (in the limits of experimental error). This is possible only in the case that in the experiments of the first and second groups, the same chemical compound was decomposed onto the same gases CO and H₂. Meanwhile the considerable difference of the ratio CO₂/CO in the experiments of the first and second groups is the result of the additional chemical reaction (Eq. 19).

Exactly this mechanism of the electrolyte decomposition at lithium-ion cells cycling is proposed by us based on the electrochemical and chemical reactions (Eqs. 7, 13–20). Thus, this experimental result is one more confirmation of the proposed mechanism of the electrolyte decomposition during cells cycling.

Third, let us consider more in detail the experimental ratio CO₂/CO = 2.4. It is possible to make an assumption that in this group of experiments, the electrolyte decomposition runs on the oxygen-type mechanism (Eqs. 17–20). Therefore, according to the oxygen-type mechanism of the electrolyte decomposition (Eqs. 17–20), the ratio CO₂/CO must be equal to two.

As a result of the experiments (Fig.3) for the four cycles of cell charge/discharge, the carbon dioxide was released in the amount of 28.6μmol m⁻²_SLP30, while the сarbon monoxide of 12μmol m⁻²_SLP30. The сarbon monoxide is released only due to the electrolyte decomposition (Eqs. 13–20). However, the gas CO₂ is released not only due to the electrolyte decomposition but instead also due to the decomposition of Li₂CO₃. (It is released in the amount of 1.5·4/2 = 3μmol m⁻²_SLP30 over the period of four cycles). Besides, a part of the released on cathode CO is transformed to CO₂ due to the residual moisture removal from the electrolyte according to the formula (Eq. 19) and maybe (Eq. 8). In the sections (see description of the first and second group of experiments (Figs. 2–3)), it was shown that thus additionally, the carbon dioxide is generated in the amount of 0.9μmol m⁻²_SLP30. Hence, only due to the electrolyte decomposition, there was released the carbon dioxide in the amount of 28.6-3-0.9 = 24.7μmol m⁻²_SLP30.

Thus only due to the electrolyte decomposition, we obtain the following value for the ratioя: CO₂/CO≈2.058. This value coincides with the theoretical value 2.0 in the limits of experimental relative error 3%.

Hence, in this group of experiments, the electrolyte decomposition runs only on the oxygen-type mechanism (Eqs. 17–20). To the same conclusion also, it is possible to come via analysis of the results of the experiments given in the Fig.3. After reaching the voltage 4.57V, the release starts of atomic oxygen from cathode; and in this case, the oxygen-type mechanism of the electrolyte decomposition is realized (Eqs. 17–20). Besides, also the released oxygen (according to Fig.3) enters into the cell head space (Fig.1). And within the following cycle from beginning of cell charge and up to the voltage 4.57V, the oxygen is consumed from the cell head space (Fig.3). Hence, again the oxygen is adsorbed on cathode (Fig.1). Then it dissociates on atoms and takes part in the electrolyte decomposition on the oxygen-type mechanism (Eqs. 17–20). Thus, over the period of the entire cycle of cell charge (Fig.1), the oxygen-type mechanism of electrolyte decomposition takes place.

The main purpose of this section is to explain the results of the third group experiments (Fig.4) quantitatively based on the established mechanism of electrolyte decomposition (Eqs. 7, 13–20).

Let us give consideration to cell cycling up to the upper limit voltage of 4.8V at 60°C (Fig.4). The amount of the released gases CO₂+CO and H₂ for four cycles grew 1.3 times as compared to the second group of experiments (Figs. 3 and 4). Hence, also the electrolyte decomposition rate grew 1.3 times. However, as it was shown (see the section Results of cells cycling in voltages range 2.6-4.8V at 60°C), at temperature rise up to 60°C, the diffusion rate of oxygen from cathode at charge stays approximately the same as in the second group of experiments. As a result, the share of CO₂sharply falls in the general balance of the gases CO and CO₂, so the ratio CO₂/CO becomes equal to 1.2.

From knowing the mechanism of the electrolyte decomposition (Eqs. 13–20), it is possible to calculate this ratio purely theoretically. According to Figs. 3 and 4, the general ratio of the released gases (CO+CO₂) in the third and second groups of experiments is equal to (CO+CO₂)₃/(CO+CO₂)₂≈1.3. The amount of the released gas CO₂ in these groups of experiments is the same (as the diffusion rate of oxygen from cathode at charge is the same (see description of the third group of experiments (Figs. 4)), i.e. (CO₂)₃ = (CO₂)₂. From theses two equations, we obtain the ratio as follows: (CO₂/CO)₃ = 1/(0.3+1.3/((CO₂/CO)₂))≈1.2 as (CO₂/CO)₂ = 2.4 (see the section Results of cells cycling in voltages range 2.6-4.8V at 25°C). Thus, the obtained calculated value for the ratio (CO₂/CO)₃ coincides with the direct experimental value 1.2 (see the section Results of cells cycling in voltages range 2.6-4.8V at 60°C).

Now let us give consideration to the dilemma, on what mechanism the electrolyte is decomposed at the temperature 60°C. The amount of the released gas CO₂ in the experiments of the second and third groups is approximately the same (Figs. 3 and 4) as at temperature rise up to 60°C, the diffusion rate of oxygen from cathode at charge stays approximately the same as at the temperature 25°C (see the section Results of cells cycling in voltages range 2.6-4.8V at 60°C). However, the amount of the released gas CO grows sharply (Figs. 3 and 4). This is possible only due to the direct mechanism of the electrolyte decomposition (Eqs. 14 and 15). According to the Figs. 3 and 4, at the temperature rise up to 60°C, the amount of the released gas CO rises approximately twice. In the second group of experiments, the gas CO was generated only due to the oxygen-type mechanism of the electrolyte decomposition. In the third group of experiments, approximately the same amount of CO is generated due to direct mechanism of the electrolyte decomposition. Hence at the cell temperature 60°C, contributions of both mechanisms into the process of the electrolyte decomposition are approximately the same. The conducted studies show that with cell temperature rise, the probability of the direct mechanism of electrolyte decomposition (Eqs. 14 and 15) increases drastically.

Additionally let us estimate other ratios for the gases released due to the electrolyte decomposition. First of all, let us evaluate the ratio of the released gases (CO₂+CO)/H₂ based on the experimental data given in the Fig.4.

According to the Fig.4 (up to the charge voltage 4.0V in the fourth cycle), the hydrogen was released in the amount of 30.1μmol m⁻²_SLP30, while the gases mixture CO+CO₂was released in the amount of 44.8μmol m⁻²_SLP30. Besides, the amount of the hydrogen released due to the decomposition of residual moisture in this group of experiments is the same as in the first and second groups of experiments (2.4μmol m⁻²_SLP30) because the same electrolyte was used. Hence, only due to the electrolyte decomposition, the hydrogen was released in the amount of 30.1-2.4 = 27.7μmol m⁻²_SLP30.

Now let us estimate the amount of the released gases CO+CO₂ (up to the charge voltage 4.0V in the fourth cycle) only due to the electrolyte decomposition. As it was shown, at the temperature rise up to 60°C, the Li₂CO₃solubility does not change much (see the section Results of cells cycling in voltages range 2.6-4.8V at 60°C). Hence, the amount of the gas CO₂ (over the period of first three cycles) released due to the Li₂CO₃ electrochemical decomposition (Eq. 5) is supposed to be approximately the same as in the first and second groups of experiments, i.e. 1.5·3/2 = 2.25μmol m⁻²_SLP30. Besides, the amount of CO₂ released due to the residual moisture removal is also approximately the same (0.9μmol m⁻²_SLP30) because the same electrolyte was used. Hence, only due to the electrolyte decomposition, the gases mixture CO₂+CO was released in the amount of 44.8-2.25-0.9 = 41.65μmol m⁻²_SLP30.

As a result, for this ratio, we obtain the experimental value (CO₂+CO)/H₂ = 1.504, which matches very well to the theoretical value 1.5 (The relative error is less than 1%). Hence, also in this group of experiments, the proposed mechanism of the electrolyte decomposition (Eqs. 7, 13–20) shows a very good correspondence between the experimental data and the theoretical calculations.

Second, let us estimate ratios for the gases released due to the electrolyte decomposition in the third and second groups of experiments. With due account of said above, in the third and second groups of experiments only due to the electrolyte decomposition, the ratio for the mixture of CO+CO₂will be equal to (CO+CO₂)₃/(CO+CO₂)₂ = 1.38, while the ratio for the amount of releasing hydrogen will be equal to (H₂)₃/(H₂)₂ = 1.37. The synchronous rise of the gases amounts of CO₂+CO and H₂ is a strong indication that in the experiments of the second group (Fig.3) and the third group (Fig.4), the decomposition runs of the same chemical compound and onto the same gases CO and H₂. Meanwhile the considerable difference of the ratio CO₂/CO in the experiments of the second and third group is the result of the additional chemical reaction (Eq. 19).

From the discussion of experimental results in the last three sections, we can draw the following conclusions about the overall mechanism of gases generation under different cycling conditions.

• (1)

  From a comparison between the electrochemical reactions (Eqs. 17 and 18) and (Eqs. 14 and 15), it becomes evident that running of the electrochemical reactions (Eqs. 17 and 18) is more energetically profitable than running of the electrochemical reactions (Eqs. 14 and 15). Thus in the case of the atomic oxygen release from the cathode (at charge voltage over 4.57V), oxygen-type mechanism of the electrolyte decomposition must take place, i.e. the reactions (Eqs. 17–20). This mechanism of electrolyte decomposition and gases generation was in the second group of experiments (see the section Discussion of second group of experiments (Fig.3)).

• (2)

  The direct mechanism of electrolyte decomposition (Eqs. 14 and 15) take place when there is not enough atomic oxygen at the cathode. For example, in the first group of experiments when cell cycling in the voltage range 2.6-4.2V, atomic oxygen is not released from the cathode. In this case, the direct mechanism of electrolyte decomposition makes a large contribution to the total amount of released gases (see the section Discussion of the first group of experiments (Fig.2)).

• (3)

  In the last section, it was shown that with increasing cell temperature, the probability of the direct mechanism of electrolyte decomposition (Eqs. 14 and 15) increases. At the same time, the velocity of the atomic oxygen diffusion from the cathode does not change much at a cell temperature growth – from 25°C to 60°C. Thus, when the cell temperature rises above 25°C, both the oxygen-type mechanism of the electrolyte decomposition (Eqs. 17–20) and the direct mechanism of electrolyte decomposition (Eqs. 14 and 15) begin to work. Moreover, at the cell temperature 60°C, contributions of both mechanisms into the process of the electrolyte decomposition are approximately the same (see the section Discussion of third group of experiments (Fig.4)).

• (4)

  In the papers,^{20,42,43,46–48} it was proved that at voltages higher than 4.0V on the cathode, there runs oxidation of Li₂CO₃with CO₂ evolution. Moreover, a half of CO₂ amount is generated due to the electrochemical decomposition of Li₂CO₃ (Eq. 5), while the second half of CO₂ amount is generated due to the electrolyte decomposition with CO subsequent oxidation on cathode under action of the atomic oxygen from the reaction (Eqs. 5), i.e. it is generated on the oxygen-type mechanism of the electrolyte decomposition (Eqs. 17–20). Simplified, this mechanism of decomposition of Li₂CO₃ and CO₂generation is represented by the reaction (Eq. 7). This CO₂generation mechanism is present in all the experiments performed. It should be noted that in this mechanism only half of the CO₂ is generated due to the electrolyte decomposition.

At present, there is no generally accepted mechanism that could quantitatively explain experiments (Figs. 2–5 and (Ref.20,21,28)) in which gases are generated during the cycling of lithium-ion cells.

In a number of papers,^20,21 various mechanisms of electrolyte decomposition and various qualitative schemes for the generation of gases were proposed. However, these schemes could not quantitatively explain the resulting gases and the ratio between them.

The mechanism of electrolyte decomposition during cells cycling, established in this paper, has several advantages.

First, electrochemical reactions (Eqs. 7, 13–20) underlying the established mechanism of electrolyte decomposition, for the first time allowed to quantitatively explain the resulting gases and the ratio between them. These studies are done in the last three sections. Therefore, the established mechanism of electrolyte decomposition during cells cycling is of great theoretical importance for understanding the electrochemical processes inside the cells during their operation.

Secondly, all electrochemical reactions leading to generation of gases and other side products in the process of cells cycling lead to the degradation and aging of cells.^14–21 In addition, the gases evolution in the lithium-ion batteries is a serious problem; it is especially so in the case of their work under high voltages and temperatures.^22,23 The established mechanism of electrolyte decomposition for the first time provides a quantitatively correct explanation of these undesirable phenomena. Only the knowledge of the electrochemical mechanism of any process allows them to manage optimally. Consequently, the established mechanism of electrolyte decomposition (Eqs. 7, 13−20) can be the basis for research to reduce the processes of degradation and increase the service life of cells.

Third, in paper,⁷⁵ it was experimentally proved that when cells cycling, hydrogen accumulates in the anode, and therefore the probability of thermal runaway in lithium-ion cells sharply increases. The established mechanism of electrolyte decomposition (Eqs. 14–20) explains the reason for the evolution of hydrogen at the anode and its accumulation in the anode (Fig.5). Knowing the causes and mechanism of hydrogen accumulation, it is possible to effectively control such a dangerous phenomenon as thermal runaway and therefore improve the safety of lithium-ion cells operation.

Fourth, in a number of papers,^20,21,28,34 it was experimentally established that some gases (H₂, C₂H₄) are released during both cell charge and discharge, i.e. in the potential independent manner. Within the framework of the established mechanism of electrolyte decomposition, it was experimentally proved for the first time that a potential-independent H₂ evolution is a consequence of its adsorption in pores of powdered graphite on anode (Fig.5).

Undoubtedly, the proposed mechanism of the electrolyte decomposition (Eqs. 7, 13–20) at cycling of lithium-ion cells requires further investigations both experimental and theoretical. Notwithstanding, it allows explaining quantitatively all the experimental data available to the date.

N. E. Galushkin 0000-0002-1613-8659

D. N. Galushkin 0000-0001-8261-6527