Yin Zhanga,
Zhen-Bo Wang*a,
Min Nieb,
Fu-Da Yua,
Yun-Fei Xiaa,
Bao-Sheng Liua,
Yuan Xuea,
Li-Li Zhenga and
Jin Wuc
aMIIT Key Laboratory of Critical Materials Technology for New Energy Conversion and Storage, School of Chemistry and Chemical Engineering, Harbin Institute of Technology, No. 92 West-Da Zhi Street, Harbin, 150001 China. E-mail: wangzhb@hit.edu.cn; Fax: +86-451-86418616; Tel: +86-451-86417853
bCollege of Chemical and Chemical Engineering, Harbin Normal University, Harbin, Heilongjiang 150025, China
cXi’an Huijie Industrial Co., Ltd., Xi’an, 710116 China
First published on 30th June 2016
Electrode materials with high tap densities and high specific volumetric energies are the key to large-scale industrial applications for the lithium ion battery industry, which faces huge challenges. LiNi0.5Co0.2Mn0.3O2 cathode materials with different particle sizes are used as the raw materials to study the effect of the mass ratio of mixed materials on the tap density and electrochemical performance of mixed materials in this work. Physical and electrochemical characterizations demonstrate that the tap density of mixed powders with different particle sizes is higher than those of materials with a single particle size. The tap density of as-prepared material has a decreasing trend with the increase of the ratio of 9 μm sized particle in the materials. The highest tap density among all of the kinds of materials reaches up to 2.66 g cm−3. Besides, the mixed material with a mass ratio of 7:
2
:
1 has a bigger specific surface area and it presents better cycle behaviors and rate capability than other materials. The specific volumetric capacity of this mixed sample reaches up to 394.3 mA h cm−3 with 1C rate charge/discharge, and it has improvements of 8.5%, 22.2% and 40.6% over any single particle size of 9 μm, 6 μm and 3 μm, respectively, which contributes to the industrial production of Li–Ni–Co–Mn–O cathode materials for lithium ion batteries.
It is well known that low tap density, which results in low volume energy density, is one of the key factors in limiting the applications and prospects of LIBs. Therefore, reducing the level of volume energy density will augment the difficulties for industrialization and commercialization. Therefore, since many researchers make efforts to improve the tap density of anode materials,8–10 it is indispensable to raise the tap density and volumetric specific energy of cathode materials for matching with the anode when they are assembled into full cells.
Many groups do a large amount of research on enhancing the tap density and volumetric specific energy of cathode materials, including spinel,11,12 olivine13,14 and layered15,16 electrodes. LiCoO2 materials are common electrodes which are widely used in lots of fields at present. However, there are some defects such as their high cost and low environmental friendliness owing to a lot of cobalt elements that limit their applications.17,18 LiNixCoyMn1−x−yO2 (NCM) materials, which were first put forward by Ohzuku in 2001,19 with an α-NaFeO2 layered structure, have been considered to be promising cathode materials to replace LiCoO2 for LIBs due to their intrinsic characteristics including high capacity and cycling stability, relatively lower cost and more safety than commercialized LiCoO2 materials.20–24 Among them, LiNi0.5Co0.2Mn0.3O2 (hereafter abbreviated as NCM523) has attracted much attention because of the improvement of the specific capacity and the reduction of cost contributed by higher Ni and lower Co content.25–27 Some research focuses on improving NCM conductivity in order to enhance the rate performance and then to meet the commercial requirements for NCM523.28–30 But unfortunately, the improving rate performance may cause the decrease of volume energy density.28 Thus, it still limits the application as mentioned above. To solve this problem, more works focus on the single particles to improve tap density, while their methods are all complicated and difficult to reproduce, and so are not suitable for large-scale industrial production. For instance, quite a number of researchers devote attention to the methods of synthesis,31–35 such as using a co-precipitation method in which all conditions are difficultly controlled simultaneously,32,34 or a complicated eutectic molten-salt method.31 And the tap densities obtained by this research reach the range of 1.72–2.89 g cm−3. Others adopt means of doping the bulk phase with Na,36 B and F,37 Cr,38 using co-precipitation methods making tap densities reach the range of 1.71–2.48 g cm−3. Therefore, it is still a challenge to obtain Ni-based cathode materials with both high electrochemical performance and high tap densities using a relatively easy method because the electrochemical properties, especially rate capability, could be improved at the expense of volumetric energy density as mentioned above, which is dependent on the tap density of the material.
In short, a high tap density is expected for cathode materials to obtain a high volumetric energy density.39,40 Thus, the volume of batteries could be smaller for the application of commercial lithium ion batteries due to the higher tap density.36 In this work, we mixed NCM523 cathode materials with different dimensions in certain weight ratios to improve tap density. Meanwhile the commercialized NCM523 with single particle size was used as the contrast sample. The value of specific volumetric capacity was used as the primary factor with consideration to estimate the electrochemical properties of all of the as-prepared samples for getting higher volumetric energy density compared with the present NCM523 material that few references mentioned as far as we know. Above all, this method in our work can obviously enhance the specific volumetric capacity of materials, and is much easier than any other reported in the literature in terms of improving the tap density and specific volumetric capacity to our knowledge, and it can be replicated easily. As a result, our method is suitable for industrialized production to improve the tap density and specific volumetric capacity due to the simple preparation process compared to the literature reported above. And more meaningful is that we believe this method may popularize other spherical or spherical-like electrode materials to meet the demand of higher tap density and volume energy density in industrial production.
Fig. 1 shows a triangle model of the weight ratios of NCM523 materials mixed with different particle sizes (9 μm, 6 μm and 3 μm) for tap density measurements. The tap densities of materials mixed with three kinds of particle sizes in various weight ratios are listed in Table S1.†
It can be seen in Table S1† that the tap density of the materials has an increasing trend with the weight ratio increase of bigger sized (D9 and D6) materials, illustrating the fact that the materials with the diameters of 9 μm will mainly contribute to a higher tap density of the mixed samples. However, sample 22 which was composed of only small size (D3) material has the lowest tap density of 1.50 g cm−3. Fig. 2 indicates the trends of tap density according to Table S1,† changing with different particle sizes.
Fig. 2 shows the curves for tap density changing with different particle sizes at various weight ratios. Fig. 2(a) is the tap density of the single samples D9, D6 and D3. Obviously, the tap density decreases with the decreasing of particle size, because smaller particles can move to the voids among bigger ones, thus sample D3 does not have enough voids for other particles smaller than itself to move into, due to the size distributions of the three samples obeying the log normal distribution law, which means that D9 has a wider distribution range than that of D3 according to our previous research.41 Besides, another reason is that the smaller size the sample has, the greater its hygroscopicity. Therefore, the powder flowability of D3 is weaker than that of D9. Fig. 2(b)–(d) indicate the influence of tap density by comparing other particle size samples based on a certain proportion of one particle size sample remaining unchanged. Namely, they are the content of D3 or/and D6 in samples with different weight ratios of particle sizes. And the curves in Fig. 2(b) and (c) indicate that an appropriate mixing of D9 and D6 (or/and D3) will achieve an increase in the tap density though D3 will reduce the tap density of mixed samples because of its extremely low tap density (1.50 g cm−3) as shown in Table S1.† Though it is not easy to find out that the tap density of samples increases obviously with the increase of the percentage of D6 in Fig. 2(d), especially from 10% to 40%, it also can be seen from Table S1† that there is a little rise indeed.
To sum up, samples will have a higher tap density as long as there is mixing of relatively smaller sized particles into bigger ones, as suggested in Fig. 2. However, the materials with lower tap densities could not meet the requirements for industrial production due to their worse volume energy density and lower economic benefit. Therefore, among all of the sampled points, we pick out six samples (red points in Fig. 1, where material numbers of 2, 3, 5, 6, 7 and 8 are named as M2, M3, M5, M6, M7 and M8, respectively) to conduct other characterizations as follows, which have higher tap densities (>2.50 g cm−3), making them an advantage in volume capacity compared with other samples.
The uniformity of the mixed samples could be roughly estimated using SEM images, and the morphological characterizations of M6 and M7 samples mentioned above are exhibited in Fig. 3, while SEM images of the other samples are shown in Fig. S1.† Furthermore, details of the particle size distribution parameters are shown in Table S2.†
From Fig. 3, it can be believed that the quasi-spherical configuration of the material is not destroyed after the grind process. Also, materials mixed with different particle sizes are all well-distributed in the view fields on the micron scale, meaning that all of the samples are mixed uniformly by different weight ratio mixing. Fig. 3(a) and (c) are the images of M6, which has a weight ratio of 7:
3
:
0 (D9
:
D6
:
D3), while Fig. 3(b) and (d) are the images of M7, which has a weight ratio of 7
:
2
:
1 (D9
:
D6
:
D3) for comparison. As a result, more interspaces have been taken up by smaller particle size of the D3 material, leading to the fact that M7 has a higher tap density than that of M6. BET data are listed in Table 1, and it can also indicate that the specific surface area is expanding with the increasing content of smaller particle sized material.
Sample | M8 | M7 | M5 | M6 | M3 | M2 |
Specific surface area (m2 g−1) | 0.3500 | 0.3277 | 0.3196 | 0.2860 | 0.2639 | 0.2507 |
The initial charge–discharge curves of the six samples with a rate of 0.2C are displayed in Fig. 4, and more related details are revealed in Table 2. It shows that M7 performs with the highest discharge specific capacity of 160.2 mA h g−1 with the highest coulombic efficiency of 85.3%, while the discharge specific capacities of M8, M2, M3, M6 and M5 reach to 158.1, 153.5, 151.4, 151.1 and 150.5 mA h g−1, respectively. In addition, the initial charge/discharge capacity has a decreasing trend with the increasing content of D6, in particular, M2, M3 and M6 demonstrate this obviously compared with other samples even though the coulombic efficiency is not affected with some great extent.
Fig. 5 shows the cycle behaviors and rate capabilities of the mixed samples, and more relevant details are revealed in Tables S3 and S4.† Fig. 5(a) and (b) show the cyclic curves of both the specific gravimetric capacity and specific volumetric capacity of the samples, while Fig. 5(c) and (d) show the rate capability of both the specific gravimetric capacity and specific volumetric capacity of the samples. It can be seen from Fig. 5(a) that the discharge capacities of the samples increase at first and then are steady with a little decrease due to the increasing content of D9. Therefore, we choose the average capacity of the first 10 cycles to evaluate the performance of the mixed samples as follows. Then, the specific volumetric capacity can be calculated by multiplying the specific gravimetric capacity by the tap density of each sample, and the results are shown in Fig. 5(b). It can be seen that the average specific volumetric capacities of both M7 and M2 for the first 10 cycles at 1C reach 394.3 mA h cm−3 which is higher than those of the other samples due to the higher tap densities they have. The average specific volumetric capacities of the last 10 cycles for M7 and M2 achieve 379.4 and 370.3 mA h g−1, with the capacity retentions of 96.2% and 96.4%, respectively. Fig. 5(c) and (d) show the rate capability of the mixed samples, and the charge–discharge process is performed with the same rate between 3.0 and 4.3 V for each 10 cycles. It demonstrates that M7 has a better rate capability than those of the samples shown in Table S4,† and its average volumetric capacities at 1C, 2C, 3C and 5C reach up to 354.9, 339.7, 325.5 and 314.6 mA h cm−3, respectively, and when it returns to 1C for the last 10 cycles, the capacity returns as well. The capacity retentions (5C/1C) of all of the samples stay at over 80%. However, M7 expresses the best rate capability with a retention of 88.7% for 5C/1C. The outstanding rate capability of M7 may be attributed to the higher content of bigger particle size materials like D9 and D6, resulting in the structural stability of the material during charge–discharge processes. Furthermore, the electrochemistry properties of the single size materials (D9, D6 and D3) are compared with the mixed sample M7 as follows, which has a better performance than that of others. Furthermore, the more visual graphical representation is exhibited in Fig. S2.†
The electrochemical properties of mixed sample M7 compared with the materials of single particle size (D9, D6 and D3) are shown in Fig. 6. Fig. 6(a) and (b) show the specific gravimetric and volume capacities of the samples while Fig. 6(c) and (d) show the rate capability of the samples. It is obvious that the specific volume capacity of M7 reaches up to 394.3 mA h cm−3, which is much higher than those of the single particle size materials due to its high tap density as shown in Fig. 7, even though it has the lower specific gravimetric capacity. Furthermore, from Fig. 6(b), it can be seen that M7, which is mixed with different particle sizes, is improved by 8.5%, 22.2% and 40.6% over samples D9, D6 and D3, respectively. Besides, the rate performance of M7 appears better than those of D9, D6 and D3 with a capacity retention for 5C/1C of 88.1%, which is higher than those of other samples according to the details shown in Tables S5 and S6.† However, it is unfortunate that the capacity retention of sample M7 is a little worse than that of D9 after 100 cycles as shown in Fig. 6(a) and (b). One of the possible reasons for this might be that electrolyte contacting is a little harder among the bigger sized particles, which results in a relatively lower capacity during about the first 10 cycles, and then the capacity rises slowly to a relatively stable status after about 10–20 cycles, because it needs a certain time for the soak between electrolyte and active materials. As a result, the capacity retention of sample D9 is more than 100%. Another reason is that the cycling stability of NCM materials has a bigger fluctuation as the temperature changes, leading to the line type of the cycle performance presenting a kind of wave. The temperature is not stable absolutely due to the restriction of the testing environment. Even so, the specific volume capacity of M7 is superior to that of the single particle size materials.
The reason for the better rate performance of M7 may be the improved speed of electron and lithium ion transport due to the smaller sized materials filling into the interspaces among particles, so that the conductive additive will contact with active materials much more closely than in samples with single size particles, as shown in Fig. 8, which is the SEM of electrode cross-section drawn before cycling, and the magnified figure in the red frame is shown in the lower left quarter of the whole picture. Also, Fig. 8 indicates that there is no significant effect on the tap density when they are made into a slurry due to its uniform mixing. Fig. 9 shows a schematic diagram of the path of electrical conduction for both single particle size materials and mixed samples. As it can be seen from Fig. 8 and 9, single particle size materials are less compact than mixed samples, which leads to the invalidation of conductive agents. Thus, that may be one of the reasons why the mixed sample has a better electrochemical performance, especially its rate property. As a result, the electrochemical performance of the mixed sample is better than any single particle size sample especially for their specific volume capacities, which is one of the most important targets for industrial production. Besides, this method to improve tap density and specific volume capacity is easier and more effective than some other works listed in Table 3, so that it can be used widely in large-scale industrial production.
![]() | ||
Fig. 9 Schematic diagram of the electrical conduction path for both single particle size materials and mixed samples. |
Sample | Method | TDa (g cm−3) | VRb (V) | SVCc (mA h cm−3) | Ref. |
---|---|---|---|---|---|
a Tap density.b Voltage range.c Specific volumetric capacity (calculated by author according to TD and VR). | |||||
Li[Ni1/3Co1/3Mn1/3]O2 | Hydroxide co-precipitation | 2.39 | 2.8–4.3 | 380.0 (at 20 mA g−1) | 44 |
Li[Ni1/3Co1/3Mn1/3]O2 | Hydroxide co-precipitation | 2.56 | 2.8–4.3 | 426.5 (at 32 mA g−1) | 45 |
Li[Ni1/3Co1/3Mn1/3]O2 | Two-step drying | 2.95 | 3.0–4.3 | 492.7 (at 0.2C) | 46 |
Li[Ni1/3Co1/3Mn1/3]O2 | Carbonate co-precipitation | 2.19 | 2.8–4.2 | 348.2 (at 20 mA g−1) | 47 |
Li[Ni1/3Co1/3Mn1/3]O2 | Hydroxide co-precipitation | 2.38 | 3.0–4.3 | 397.4 (at 16 mA g−1) | 48 |
LiNi0.8Co0.15Mn0.05O2 | Hydroxide co-precipitation | 2.72 | 3.0–4.3 | 478.7 (at 1C) | 49 |
LiNi0.7Co0.15Mn0.15O2 | Hydroxide co-precipitation | 2.37 | 3.0–4.3 | 438.9 (at 0.2C) | 50 |
LiNi0.6Co0.2Mn0.2O2 | Hydroxide co-precipitation | 2.59 | 2.8–4.3 | 445.7 (at 1C) | 51 |
Li[Ni1/3Co1/3Mn1/3]O2 | Eutectic molten-salt | 2.89 | 3.0–4.3 | 462.4 (at 0.2C) | 31 |
Li[Ni0.5Co0.2Mn0.3]O2 | Hydroxide co-precipitation | 2.68 | 3.0–4.3 | 442.2 (at 0.2C) | 32 |
Carbonate co-precipitation | 2.35 | 364.3 (at 0.2C) | |||
Li[Ni0.6Co0.2Mn0.2]O2 | Hydroxide co-precipitation | 2.32 | 2.8–4.3 | 399.7 (at 1C) | 34 |
Li0.97Na0.03Ni0.5Co0.2Mn0.3O2 | Hydroxide co-precipitation | 2.17 | 3.0–4.6 | 353.9 (at 1C) | 36 |
Li[(Ni1/3Co1/3Mn1/3)0.98Al0.02]O2 | Hydroxide co-precipitation | 2.12 | 3.0–4.4 | 332.8 (at 1C) | 37 |
Li[(Ni1/3Co1/3Mn1/3)0.96Al0.02B0.02]O1.98F0.02 | 2.48 | 391.8 (at 1C) | |||
LiNi0.35Co0.2Cr0.1Mn0.35O2 | Hydroxide co-precipitation | 3.10 | 2.5–4.5 | 511.5 (at 20 mA g−1) | 38 |
Li[Ni0.5Co0.2Mn0.3]O2 | Simple homogeneous hybrid | 2.66 | 3.0–4.3 | 394.3 (at 1C) | Our work |
Cyclic voltammetry (CV) measurements of the mixed material M7 compared to D9 are presented in Fig. 10 and the details of the other mixed samples are shown in Table S7.† It is well known that there are three transition metal elements including Ni, Co, and Mn with different valences of +2, +3 and +4, respectively in NCM cathode materials. And Ni2+ ions and Co3+ ions act as the electrochemically active elements to conduct redox reactions while Mn4+ ions remain inert during the process of charge–discharge.52–54 The redox potential of a sample can be evaluated from the CV curves. The anodic peaks at around 3.8 V within the potential range of 3.0–4.3 V in Fig. 10 for D9 and M7 correspond to the oxidation process of Ni ions from Ni2+ to Ni3+ and Co ions from Co3+ to Co4+, accompanying the process of Li+ de-intercalation. The cathodic peak at around 3.7 V presents the reverse electrochemical process of Li+ intercalation. It can be seen from Fig. 10(a) that M7 has a lower potential at the beginning of the peak than that of D9. And the oxidation peak potential of M7 is 3.825 V, which is lower than D9 (3.851 V) due to the bigger specific surface area of M7 for contact with electrolyte. Besides, all of the mixed samples have a lower potential at the beginning of the peak appearing compared to D9 as shown in Table S7.† However, Fig. 10(b) shows the curves of the samples after 100 cycles. Unfortunately, the cycle performance of D9 is better than M7 after 100 cycles according to the decrease of their peak potential differences (ΔE). However, the result conforms to the capacity of D9, as the capacity retention is more than 100% after the charge–discharge process for 100 cycles in Fig. 6(a) and Table S5† due to difficulties in electrolyte contacting among bigger size particles. In spite of this, the better performance of the mixed samples could remedy the deficiencies of single particle size materials especially for industrial production. Moreover, the rate property is also good, informed by the electrochemical impedance spectroscopy (EIS) data.
Fig. 11 shows the electrochemical impedance spectroscopy (EIS) curves of mixed material M7 and single size sample D9 before the charge–discharge process, and the parameters of equivalent circuit for them are listed in Table 4. It can be seen from Fig. 11(a) that the curves all consist of three semicircles and an oblique line at high, medium-high, intermediate and low frequencies, respectively. The intercept on the real axis and the first semicircle stands for the ohmic resistance (Rs) and the resistance of lithium ion diffusion through the solid electrolyte interphase film (Rsei). The second semicircle represents the interface electronic resistance (Re) of active materials, while the third semicircle represents the charge transfer resistance (Rct), which is the important factor to investigate electrochemical performance, especially the rate property of materials. In addition, the oblique line at low frequency is the impedance for lithium-ion diffusion inside material bulk.55–57 Fig. 11 and Table 4 illustrate that M7 has a smaller value of Rct (7.513 Ω) than that of D9 (15.12 Ω), which is the reason for its better rate performance. Fig. 11(b) shows the relationships between Z′ and ω−1/2 of samples D9 and M7. Here, ω is angular frequency, which can be obtained by eqn (1). Furthermore, the Li-ion diffusion coefficient of materials can be obtained according to the formula (2) as shown below.58
ω = 2πf | (1) |
![]() | (2) |
![]() | ||
Fig. 11 EIS curves of D9 and M7 before the charge–discharge process: (a) Nyquist plots of D9 and M7; (b) the relationships between Z′ and ω−1/2 for D9 and M7. |
Sample | Rs (Ω) | Rsei (Ω) | Re (Ω) | Rct (Ω) |
---|---|---|---|---|
D9 | 6.767 | 6.959 | 6.434 | 15.12 |
M7 | 5.152 | 8.878 | 8.278 | 7.513 |
In formula (2), there are four constants of R, T, n and F that represent the gas constant, temperature, number of electrons in the reaction and Faraday constant, respectively. Meanwhile, the surface area (A) and the concentration of Li+ (C) are of no difference in the same system. Therefore, the Warburg coefficient (σ), which is the slope in Fig. 11(b) mentioned above, becomes the only factor to affect the value of DLi+. Also, it can be seen from formula (2) that DLi+ has a negative correlation with σ, namely, the bigger slope it shows, the smaller DLi+ the sample has, and vice versa. Hence, it is shown in Fig. 11(b) that the slope value of M7 is 3.79293, which is smaller than that of D9, leading to the bigger DLi+ of M7 than that of D9. As a result, it can also illustrate why sample M7 has a better rate performance than that of D9, which is in agreement with Fig. 6(d).
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11052e |
This journal is © The Royal Society of Chemistry 2016 |