Degradation Identification of Individual Components ... - Science Direct

25 downloads 0 Views 1MB Size Report
doi: 10.1016/j.egypro.2017.03.919. Energy Procedia 105 ( 2017 ) 2698 – 2704. ScienceDirect. The 8th International Conference on Applied Energy – ICAE2016.
Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 105 (2017) 2698 – 2704

The 8th International Conference on Applied Energy – ICAE2016

Degradation identification of individual components in the LiyNi1/3Co1/3Mn1/3O2-LiyMn2O4 blended cathode for large format lithium ion battery Dongsheng RENa, Languang LUa,b, Minggao OUYANGa,*, Xuning FENGa, Jianqiu LIa,Xuebing HANa a

. State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing 100084, China. b . Collaborative Innovation Center of Electric Vehicles, Beijing 100084, China

Abstract The degradation behaviors of individual components in the Li yNi1/3Co1/3Mn1/3O2(NCM)–LiyMn2O4(LMO) blended cathode are investigated in this paper. A prognostic/mechanistic model is applied to quantitatively analyze the capacity fading mechanism of the lithium ion battery with blended cathode. NCM and LMO are considered to be connected in parallel with a same potential during cycling. Current is distributed between NCM and LMO based on the dQ/dV curves, and the capacities of NCM and LMO can thus be quantitatively identified, respectively. The identified model parameters show that both NCM and LMO suffer loss of active material (LAM), but NCM degrades faster than LMO. Loss of lithium inventory also contributes to the capacity degradation, whereas anode shows little LAM until 330 cycles. The identified capacity degradation mechanism is validated by the coin cells test data, as the capacity degradation of the coin cells assembled with the aged electrodes compared with those made of fresh electrodes are consistent with the identified results. © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2016 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). responsibility of of ICAE Selection and/or peer-reviewofunder Peer-review under responsibility the scientific committee the 8th International Conference on Applied Energy. Keywords: Lithium ion battery; Blended cathode; Capacity degradation mechanism; Prognostic/mechanistic model

Nomenclature NCM LMO LAM LLI HPPC Q V R30s

LixNi1/3Co1/3Mn1/3O2 LixMn2O4 Loss of active material Loss of lithium inventory Hybrid pulse power characterization Capacity (Ah) Voltage (V) 30s discharge internal resistance (mΩ)

R x x0 CN

yNCM yLMO y0,NCM y0,LMO

Average resistance of the battery (mΩ) Lithium concentration in LixC6 Initial value of x at the beginning of discharge Capacity of the anode (Ah) Lithium concentration in LixNi1/3Co1/3Mn1/3O2 Lithium concentration in LixMn2O4 Initial value of yNCM at the beginning of discharge Initial value of yLMO at the beginning of discharge

* Corresponding author. Tel.: +86-10-62773437; fax: +86-10-62785708. E-mail address: [email protected].

1876-6102 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 8th International Conference on Applied Energy. doi:10.1016/j.egypro.2017.03.919

Dongsheng Ren et al. / Energy Procedia 105 (2017) 2698 – 2704

tk Vsim Vp Vn I

Sampling time (s) Simulated full-cell voltage (V) Cathode potential (V) Anode potential (V) Current (A)

c CNCM CLMO k0 ∑(Li/Li+)

Mass ratio between LMO and NCM Capacity of NCM (Ah) Capacity of LMO (Ah) Initial capacity ratio between LMO and NCM Total available lithium inventory (Ah)

1. Introduction Lithium ion batteries are widely utilized as power sources for electric vehicles(EV), considering their high energy density, and extended cycle life. Recent years, blending cathode materials is becoming an applicable approach to improve the performance of the lithium ion batteries[1]. One example is the LiyNi1/3Co1/3Mn1/3O2 (NCM)-LiyMn2O4 (LMO) blended cathode. NCM exhibits high specific capacity, but suffers poor thermal stability; LMO has high thermal stability, but suffers a relatively low specific capacity[1].NCM-LMO blended cathode exhibits better safety property than NCM, and has a higher specific capacity than LMO[1]. Capacity degradation of lithium ion batteries with blended cathode can be induced by loss of lithium ion inventory (LLI) and loss of active material (LAM)[2,3]. LLI can be caused by the growth of SEI film [2], while LAM is more complicate, for NCM and LMO have different aging mechanisms[4]. NCM suffers the dissolution of transition metals (Ni, Co and Mn) and structural changes, while LMO suffers Mn dissolution at evaluated temperature [5]. The identification of battery aging mechanisms in real-life operation has been a challenging goal in battery R&D. Prognostic/mechanistic models were also applied to quantify the capacity fading mechanism in ref.[2,3,6]. However, those models cannot be applied to analyze the degradation of lithium ion batteries with blended cathode, for the two active materials in the blended cathode may fade at different rates[4]. To our best knowledge, degradation behaviors of individual components in the NCM-LMO blended cathode are still unclear. Therefore, in this research, the capacity degradation of a 20Ah pouch lithium ion battery with NCMLMO blended cathodes is investigated. A prognostic/mechanistic model is proposed to quantify the degradation of the individual components in the NCM-LMO blended cathode. NCM and LMO are considered to be connected in parallel with a same potential during cycling in the model. Current is distributed between NCM and LMO based on the dQ/dV curves, and the capacities of NCM and LMO can thus be identified separately. The identified aging mechanism is validated by the experimental results of coin cells assembled using disks of electrodes cut from the aged battery. 2. Experiments

Fig. 1 SEM image of the NCM-LMO blended cathode cut from a fresh battery

A 20Ah commercial pouch lithium ion battery with NCM-LMO blended cathode and graphite anode was tested in this study. Fig. 1 shows the SEM image of the blended cathode cut from a fresh battery. Particles made by NCM and LMO can be observed in the SEM image, respectively. Inductively coupled plasma mass spectrometry (ICP-MS) measurements helped characterize the blended cathode. The element mass proportions were Ni:Co:Mn=11:11:29.5, indicating that the mass ratio between NCM and LMO was approximately 1:1. 2.1. Reference performance test

2699

2700

Dongsheng Ren et al. / Energy Procedia 105 (2017) 2698 – 2704

Reference performance test (RPT), including capacity test and hybrid pulse power characterization (HPPC) test, was conducted on the battery at 25oC after every 30 accelerated aging tests. The battery was cycled between 2.5V and 4.2V with 0.5C current to measure the discharge capacity.

 Fig. 2 HPPC profile used to characterize the internal resistance of the battery

HPPC test was carried out to characterize the internal resistance of the battery. Prior to the pulse power sequence, the battery was fully charged to 4.2V with 0.5C current and then rested for 3h. The pulse power sequence was composed of three steps: 1) 2C discharge for 30s; 2) rest for 40s; 3) 1.5C charge for 10s, as shown in Fig. 2. The 30s discharge internal resistance of the battery can be calculated by Eqn.(1). After the pulse power sequence, the battery was discharged to 90% SOC at 0.5C, and rested for 3h before the next pulse power sequence. The battery was tested at 10% SOC decrements until reaching the cut-off voltage (2.5V).

R30 s

V1  V2 I1

(1)

2.2. Accelerated aging test Accelerated aging tests, were conducted on the battery at 50oC. The battery was first fully charged to 4.2V at 0.5C current. Then, dynamic stress test (DST) profiles (referring to [7]) were implemented repetitively to test the effect of dynamic EV driving behavior on the battery life. After 16 DST profiles, the battery was discharged to 2.5V at 0.5C current. The battery was cooled down to 25 oC after every 30 accelerated aging tests, followed by RPT to characterize the variation of capacity and internal resistance. 2.3. Coin cell test The battery was dismantled after the aging process, and electrodes disks with a diameter of 8mm were cut from the aged battery. Coin cells with NCM-LMO/Li and graphite/Li electrodes were assembled using those electrode disks after removing the active materials on one side of the current collectors. For comparison, coin cells were also prepared with electrode disks cut from a fresh battery. The coin cells were cycled under a quasi-electrochemical-equilibrium condition with a current of 0.05mA (less than 1/25C) to measure their capacities and obtain the equilibrium potential curves of the cathode and anode. The coin cells with NCM-LMO/Li electrodes were cycled in a voltage range between 3.0V and 4.3V, while those with graphite/Li electrodes were cycled between 0.05V and 1.5V. 3. Model development Here we exploit a prognostic/mechanistic model to investigate the aging behavior of the battery. Fig. 3(a) presents the physical model that captures the mechanistic performance of a lithium ion battery with NCM-LMO blended cathode. The model for the discharge voltage curve is built by Eqn.(2). At sampling time tk, the full-cell voltage Vsim(tk) is the difference of the cathode potential Vp and anode potential Vn with a potential loss of I·R, as illustrated in Fig. 3(b). I=10A is the discharge current, R is the average resistance of the battery throughout the discharge process. The anode potential Vn depends on the average

2701

Dongsheng Ren et al. / Energy Procedia 105 (2017) 2698 – 2704

lithium concentration x in LixC6. During the discharge process, x(tk) changes proportionally to current I according to Eqn.(3), where x0 is the initial value of x at the beginning of discharge, CN represents the capacity of the anode. For the NCM-LMO blended cathode, the NCM and LMO active material share a same potential during the lithiation/delithiation process given their parallel connection[8], as shown in Fig. 3(a) and Eqn.(4). The cathode potential Vp,NCM (Vp,LMO) is regarded as a function of yNCM (yLMO), which represents the stoichiometric coefficient for the lithium intercalated in NCM (LMO).

Vsim (tk ) Vp (tk )  Vn ( x(tk ))  I ˜ R x(tk )

x0  I ˜ tk / C N

Vp (tk ) Vp, NCM ( yNCM (tk )) Vp , LMO ( yLMO (tk ))

(2) (3) (4)

Fig. 3 Model of a lithium ion battery with blended cathode. (a) electrical connections in the model; (b) the prognostic/mechanistic model used for the full-cell discharge voltage curve

Fig. 4(a) shows the dQ/dV curve for the blended cathode comparing with those for the individual components. The dQ/dV value of the blended cathode is the superimposition of those of NCM and LMO. Suppose the cathode potential decreases by dVp(tk) between sampling time tk and tk+1, then the discharge capacity I ˜ (tk 1  tk ) is distributed to NCM and LMO according to Eqn.(5), where c represents the mass ratio between LMO and NCM. It should be noted that the ratio c will change during the aging process. Thus, the percentage of current I that passes through NCM can be determined by the dQ/dV curves, as shown in Eqn.(6) and Fig. 4(b). The change of yNCM can be calculated by Eqn.(7). y0,NCM is the initial value of yNCM at the beginning of discharge, CNCM is the total capacity of NCM, and Δt is the sampling interval. Besides, the capacity of LMO can be calculated by Eqn.(8), where k0 is the initial capacity ratio between LMO and NCM according to the dQ/dV curves. According to Eqn.(2)~(7), given an appropriate setting of [x0, CN, y0,NCM, CNCM, c, R], the simulated full-cell discharge voltage curve Vsim can match the measured discharge voltage curve V well, as shown in Fig. 3(b). The root mean squared error (RMSE) between the simulated and measured voltage is defined by Eqn.(9) to evaluate the fitness of the model. A smaller RMSE indicates a better degree of coincidence. Optimization method, such as genetic algorithm, can be utilized to identify the optimal set of [x0, Cn, y0,NCM, CNCM, c, R] with the fitness function of Eqn.(9).

Fig. 4 (a)dQ/dV curve for the blended cathode comparing with those for the individual components; (b) the mechanism for the current distribution within the NCM and LMO cathode connected in parallel.

2702

Dongsheng Ren et al. / Energy Procedia 105 (2017) 2698 – 2704

I ˜ (tk 1  tk )

dQNCM (tk )  dQLMO (tk ) (dQNCM / dV

P(tk )

(dQNCM / dV

V V p ( tk )

V Vp ( tk )

) ˜ dVp (tk )  c ˜ (dQLMO / dV

) / (dQNCM / dV

V Vp ( tk )

V V p ( tk )

 c ˜ dQLMO / dV

) ˜ dV p (tk ) V Vp ( tk )

)

(5) (6)

tk

yNCM (tk )

y0, NCM  ¦ P(tk ) ˜ I ˜'t / CNCM

(7)

t 0

CLMO

c ˜ k0 ˜ C NCM

n

RMSE

¦ (V (t

k

)  Vsim (tk )) 2 / n

(8) (9)

k 1

4. Results and Discussions

Fig. 5 Evolution of the discharge capacity and the 30s discharge resistance at 100%SOC during the aging process.

Fig. 5 shows the evolution of the discharge capacity and the 30s discharge resistance at 100%SOC. The discharge capacity steadily depletes from 20.14Ah to 16.99Ah after 300 cycles. Then, the capacity degradation accelerates after 330 cycle, and quickly decreases to 13.45Ah after 390 cycle. Significant increase of the 30s discharge resistance can only be found after 330 cycles.

Fig. 6 The identified results using the prognostic/mechanistic model. (a) comparison between the simulated and the experimental full-cell discharge voltage curves after different cycles. (b), (c) evolution of the identified parameters.

The prognostic/mechanistic model proposed in Sec.3 is applied to study the capacity degradation mechanism of the battery. Fig. 6 (a) compares the simulated discharge voltage curves with the experiment after different cycles. Note that the RMSEs are always less than 6mV, the simulated discharge voltage curves fit well with the experimental results, indicating that the identified model parameters can reflect the battery internal characteristics. The evolution of the identified parameters is shown in Fig. 6 (b) and (c). The change of CNCM, CLMO and CN can reflect the LAM at NCM, LMO and anode, respectively. The total available lithium inventory ∑(Li/Li+) can be calculated by Eqn. (10), where y0,LMO is the initial value of yLMO at the beginning of discharge. The decline of ∑(Li/Li+) represents LLI inside the battery. The increase of identified R reflects the increment of the internal resistance. Fig. 6 (b) shows that CNCM and

2703

Dongsheng Ren et al. / Energy Procedia 105 (2017) 2698 – 2704

CLMO both fade, implying that LAM happens at both NCM and LMO. However, NCM degrades faster than LMO. For the anode, CN shows no obvious drop until 330th cycles. The decline of ∑(Li/Li+) indicates serious LLI inside the battery. Significant rise of the identified R can be observed after the 330th cycle, consistent with the HPPC results (Fig. 5).

¦(Li / Li



)

x0CN  y0, NCM CNCM  y0, LMOCLMO

(10)

Fig. 7 The potential curves and dQ/dV curves for the fresh and aged electrodes. (a) dQ/dV curves for the cathodes; (b) potential curves for the cathodes;(c) dQ/dV curves for the anodes; (d) potential curves for the anodes. Table 1 The capacities of the fresh and aged electrodes Electrodes Cathode Anode

1# 2# 1# 2#

Capacity(mAh) Fresh Aged 1.8517 1.3600 1.9253 1.0780 2.4090 1.5839 2.4226 2.0299

Average capacity loss 35.45% 25.20%

Coin cells test is used to validate the identified capacity fading mechanism. As listed in Table 1, the capacities of the aged cathodes and anodes both degrade compared with the fresh electrodes, confirming that LAM happens at both cathode and anode. The average capacity loss of the aged cathode and anode are 35.45% and 25.20%, respectively, consistent with the identified results presented in Fig. 6(a), where LAM at the cathode and the anode are 34.97% and 20.61%, respectively. Fig. 7 shows the potential and dQ/dV curves for the fresh and aged electrodes. The changes of the peaks of the dQ/dV curves can reveal the change of electrochemical properties between the fresh and aged electrodes. Three obvious peaks can be found in the dQ/dV curves for the cathodes (Fig. 7(a)), marked as I, II and III. Peaks I and II represent the two phase transition processes of LMO, while peak III at around 3.75V indicates the phase transition process of NCM[4]. Both of the aged cathodes exhibit a decline in the height of peak III comparing with the fresh cathodes, while the height of peak I and II increase slightly, indicating a reduction of the mass ratio of NCM[4]. The decline of the mass ratio of NCM active material confirms that NCM degrades faster than LMO. In contrast to the cathodes, the potential curves and dQ/dV curves for the aged anodes show little difference from those for the fresh anodes, as shown in Fig. 7(c),(d), demonstrating that the electrochemical properties of the graphite anode stay almost unchanged. Based on the model results and the coin cell test results, the capacity degradation in the early 330 cycles mainly arises from LLI and

2704

Dongsheng Ren et al. / Energy Procedia 105 (2017) 2698 – 2704

LAM at NCM and LMO. LAM at the anode and increment of the internal resistance become evident after the 330th cycle. The good agreement between the model results and the experimental data indicates that the model proposed in this study is effective for investigating the degradation mechanism of the lithium ion battery with NCM-LMO blended cathode. Furthermore, the prognostic/mechanistic model can also help to built up dynamic capacity degradation models which can capture the degradation dynamic of li-ion batteries and be used to evaluate the longevity of a battery system under varying working conditions, as shown in our previous paper[3]. 5. Conclusions The degradation of the individual components in the NCM-LMO blended cathode is investigated in this research. The capacity degradation mechanism is quantitatively analyzed using a prognostic/ mechanistic model, which can identify the LAM at NCM and LMO in the blended cathode separately. The identified model parameters show that LAM happens at both NCM and LMO, but NCM degrades faster than LMO. LLI also contributes to the capacity degradation, whereas no obvious LAM happens at anode until the 330th cycle. Significant increase of the identified R can also be observed after the 330th cycle, consistent with the HPPC results. Coin cells with electrodes cut from the aged battery were tested to validate the identified capacity degradation mechanism. The aged cathode and anode both show significant capacity degradation compared with the fresh electrodes, indicating LAM happens at both cathode and anode. The coin cells experimental data also confirms that NCM degrades faster than LMO, as the peak III in the dQ/dV curves for the blended cathodes decreases significantly. Acknowledgement This work is funded by the National Natural Science Foundation of China (No. U1564205) References [1] Chikkannanaver. SB, Bernardi DM, Li L. A review of blended cathode materials for use in Li-ion batteries. J. Power Sources. 2014;248:91-100. [2] Han X, Ouyang M, Lu L, et al. A comparative study of commerical lithium ion battery cycle life in electrical vehicel: Aging mechanism identification. J. Power Sources. 2014;251:38-54. [3] Ouyang M, Feng X. Han X, et al. Adynamic capacity degradation model and its application condisdering varying load of a large format Li-ion battery. Applied Energy. 2016;165:48-59. [4] Stiaszny B, Ziegler JC, Krauß EE, et al. Electrochemical characterization and post-mortem analysis of aged LiMn2O4NMC/graphite lithium ion batteries part I: Cycle aging. J. Power Sources. 2014;251:439-450. [5] Wohlfahrt-Mehrens M, Vogler C, Garche J. Aging mechanisms of lithium cathode materials. Journal of Power Sources. 2004;127:58-64 [6] Ouyang M, Ren D, Lu L, Li J, Feng X, Han X, et al. Overcharge-induced capacity fading analysis for large format lithium-ion batteries with LiyNi1/3Co1/3Mn1/3O2+LiyMn2O4 composite cathode. Journal of Power Sources. 2015;279:626-35. [7] USABC. Electric Vehicle Battery Test Procedures Manual, Rev 2, 1996; [8] Albertus P, Christensen J, Newman J. Experiments on and modeling of positive electrodes with mutiple active materials for lithium-ion batteries. J. Electrochem. Soc. 2009;156:A606-A618.

Biography Dongsheng Ren received the B.E. degree and is currently pursuing PhD degree in Department of Automotive Engineering, Tsinghua University. His research focuses on the degradation behavior and safety performance of li-ion battery for electrical vehicles.