Silicon Nanoparticles - ACS Publications - American Chemical Society

Aug 30, 2017 - Argonne National Laboratory, 9700 S. Cass Avenue, Lemont, Illinois 60439, .... assembly, modeling, and prototyping (CAMP) facility at A...
0 downloads 0 Views 6MB Size
Research Article www.acsami.org

Silicon Nanoparticles: Stability in Aqueous Slurries and the Optimization of the Oxide Layer Thickness for Optimal Electrochemical Performance Linghong Zhang, Yuzi Liu, Baris Key, Stephen E. Trask, Zhenzhen Yang, and Wenquan Lu* Argonne National Laboratory, 9700 S. Cass Avenue, Lemont, Illinois 60439, United States S Supporting Information *

ABSTRACT: In this study, silicon nanoparticles are oxidized in a controlled manner to obtain different thicknesses of SiO2 layers. Their stability in aqueous slurries as well as the effect of oxide layer thickness on the electrochemical performance of the silicon anodes is evaluated. Our results show that slightly increasing the oxide layer of silicon nanoparticles significantly improves the stability of the nanoparticles in aqueous slurries and does not compromise the initial electrochemical performance of the electrodes. A careful comparison of the rate and cycle performance between 400 °C treated Si nanoparticles and pristine Si nanoparticles shows that by treating the silicon nanoparticles in air for slightly increasing the oxide layer, improvement in both rate and cycle performance can be achieved. KEYWORDS: silicon anode, surface oxide layer, stability in aqueous slurries, electrochemical performance, lithium-ion batteries the reversible capacity of the material.13 With the current trend moving toward water-based laminate-making processes due to both better electrode performance and economic reasons,14,15 the presence of the oxide layer in the silicon electrodes is inevitable. Therefore, we need to consider the Si electrode materials (especially Si nanomaterials) used with aqueous binders as Si core materials with a thin SiO2 shell and understand their stability and electrochemical performance as a whole. The purpose of this work is to understand how the oxide layer thickness affects the stability of the Si nanoparticles in aqueous slurries and furthermore to optimize the oxide layer thickness to achieve optimal initial electrochemical performance as well as cycle performance. In this article, we investigated the effect of the oxide layer on the electrochemical performance of silicon anodes by growing the oxide layer of Si nanoparticles in a controlled manner. Si nanoparticles with an average size of 80 nm were used. Lithium polyacrylate (LiPAA) with a pH of ∼7 and carbon black were mixed with the Si material to fabricate the electrodes. Brunauer−Emmett−Teller (BET) surface analysis, transmission electron microscopy (TEM), as well as Fourier-transform infrared (FTIR) spectroscopy were performed to characterize the silicon nanoparticles. Finally, the electrochemical performance of the silicon anodes with different oxide layer thicknesses

1. INTRODUCTION The fast-growing demand for high-energy-density Li-ion batteries for use in electric vehicles and portable devices urges the development of the next-generation high-energydensity electrode materials. Compared to commercial graphite anode materials, the naturally abundant silicon offers 10 times the gravimetric capacity and 3 times the volumetric capacity,1 making it a promising next-generation anode material and attracting various research in the field.1−8 When silicon is in contact with air, a native oxide layer of several nanometers is formed.9,10 As the size of the silicon particle becomes smaller, this oxide layer plays a more and more important role, as it takes up a larger volume and weight percentage. The actual thickness of the oxide layer may also vary due to the manufacturing processes and storage history. Although the use of Si nanoparticles largely mitigates the particle pulverization problem observed in micron-sized Si particles, the impact of the oxide layer on the electrochemical performance of Si nanomaterials becomes nonmarginal. Xun and co-workers reported that the presence of the oxide layer harms the initial performance of the silicon anodes in a nonaqueous system.11,12 When the oxide layer was partially etched away by HF, improved initial reversible capacity as well as decreased initial irreversible capacity loss per surface area was observed. Recently, Touidjine and co-workers reported that the native oxide layer around silicon is not thick enough to prevent the further reaction of water with the inner silicon core.13 This wet oxidation reaction forms porous silicon oxide, which is accompanied by the generation of hydrogen gas and also harms © 2017 American Chemical Society

Received: June 25, 2017 Accepted: August 30, 2017 Published: August 30, 2017 32727

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces

pouches were put into a 75 °C oven and the swelling behavior at the elevated temperature was monitored.

was analyzed and the oxide thickness was optimized on the basis of the electrochemical performance.

3. RESULTS AND DISCUSSION 3.1. Materials Characterization. Figure 1a shows the SEM image of the as-received Si nanoparticles. The particles are

2. EXPERIMENTAL SECTION 2.1. Materials and Process. Si nanoparticles (average size 80 nm) were obtained from Hydro-Québec, Inc. For a controlled oxidation process, the Si nanoparticles were placed into an air-convection oven for a fixed time period (15 h) at different temperatures (300, 400, 500, 600, and 700 °C). The samples were then cooled inside the oven before taking out for characterization and electrochemical testing. 2.2. Instrumentation. Fourier-transform infrared spectra were measured with a Thermo Fisher Nicolet iS5 FTIR spectrometer in the attenuated total reflection mode. Scanning electron microscopy measurements were performed using a Hitachi S-4700 scanning electron microscope. Transmission electron microscopy measurements were conducted using a JEOL JEM-2100F field emission transmission electron microscope. The powder specific surface area was measured by the Brunauer−Emmett−Teller (BET) N2 adsorption method. Xray photoelectron spectroscopy (XPS) measurements were conducted using a Kratos AXIS-165 Surface Analysis System with an Al source. 29 Si magic angle spinning (MAS) NMR experiments were performed at 7.02 T (300 MHz) on a Bruker Avance III spectrometer operating at a Larmor frequency of 59.63 MHz. A calibrated π/2 pulse width of 5 μs was used. The MAS spectra were acquired at a spinning speed of 20 kHz using 3.2 mm rotors with a single pulse (90°) experiment. Despite the methodological challenges and to obtain full signal saturation, that is, full quantification from slow (elemental Si)and fast-relaxing (amorphous SiO2) silicon sites simultaneously, an optimized recycle delay of 6400 s (see variable recycle experiments in Figure S9) and 36 scans were used to collect the spectra. All chemical shifts were referenced to a tetramethylsilane reference at 0 ppm. 2.3. Laminate Fabrication and Cell Assembly. A silicon/ carbon black/LiPAA mixture with a 7:1:2 weight ratio was used in this study. The mixture was first mixed using a planetary centrifugal mixer (Thinky Mixer) at 2000 rpm until the slurry became uniform without obvious aggregates and then coated onto a 10 μm thick copper foil. The laminates were first dried in a 75 °C oven for 4 h and then transferred to a 75 °C vacuum oven for overnight drying. To assemble coin cells, 15 mm diameter electrodes were punched from the Si laminates. The electrodes were first individually calendared to a porosity of ∼45% and were then dried in a vacuum oven at 150− 160 °C overnight inside the glovebox before coin cell assembly. The higher drying temperature was chosen to more effectively remove the water adsorbed on the silicon nanoparticles and the LiPAA binder without damaging the binder.16,17 A 90 wt % Gen II electrolyte (EC/ EMC = 3:7 by weight, 1.2 M LiPF6) with 10 wt % FEC was used as the electrolyte. For half-cell testing, lithium-metal foil was used as the counter electrode. For full-cell testing, LiNi0.5Co0.2Mn0.3O2 (NCM523, Toda America Inc.) was used as the cathode to obtain an N/P ratio of 1.1−1.2. The NCM523 cathode, with 90% NCM523, 5% carbon black, and 5% poly(vinylidene difluoride) binder, was fabricated by the cell assembly, modeling, and prototyping (CAMP) facility at Argonne National Laboratory. The coin cell performance was evaluated with a Maccor Series 4000 Battery Test System in a temperature-controlled chamber at 30 °C. For all cell testing, the active loading of the Si electrode is around 2 mAh/cm2. 2.4. Gas Generation Test. In the gas generation test, the slurry was made by mixing silicon/carbon black/LiPAA in a weight ratio of 7:1:2. Deionized water (18 MΩ) was added to obtain good slurry property. After the slurry was made using a planetary centrifugal mixer (Thinky mixer), 2.8−2.9 g of the slurry was immediately transferred to a pouch container that was formed to xx3450 dimensions using a pouch-forming machine (MEDIA TECH Co., Ltd). For clarification, the purpose of the pocket formed in the pouch container laminate was to provide a volume of space for the slurry to be contained for this experiment. The pouches were then vacuumed in the vacuum sealer (Busch Korea) to evacuate the air inside the pouch and finally heatsealed.18 Afterward, the pouches were placed in a fume hood and monitored for swelling behavior. After a room-temperature test, the

Figure 1. (a) Scanning electron microscopy image of silicon nanoparticles. (b) Mass change of the Si nanoparticles after heat treatment at different temperatures.

sphere shaped, with most particles in the ∼100 nm range. Some larger particles with diameters of several hundred nanometers can also be observed. The roundness and detailed feature can be seen more clearly in the TEM image shown in Figure 2a. To obtain Si nanoparticles with different oxide layer thicknesses, the Si nanoparticles were placed into an air-convection oven for a fixed time period (15 h) at different temperatures (300, 400, 500, 600, and 700 °C). The mass change of the particles was recorded after each heat treatment. Figure 1b shows the mass change of silicon particles after heat treatment at different temperatures. As the heating temperature increases, the mass change increased from 1.9% at 300 °C to 35% at 700 °C, in an exponential manner, showing that the oxidation process is kinetically controlled. We then conducted TEM to map the oxygen content of the Si nanoparticles. Figure 2b−d shows the oxygen K map of the nontreated pristine nanoparticles as well as the nanoparticles treated in air at 400 and 600 °C. The oxygen K map of the nontreated silicon particles did not show an obvious oxygenrich layer around the particles, suggesting that the thickness of the native oxide layer is within the mapping resolution of the TEM we used. Nevertheless, estimating from the TEM oxygen K map, silicon nanoparticles treated in air at 400 and 600 °C show increase in the oxide layer thickness. The oxide layer thickness increase after the heat treatment can be calculated using the weight gain shown in Figure 1b. Assuming an average 32728

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces

shell (based on TEM and X-ray diffraction findings) of the largely spherical particles. With heat treatment, an increasing trend in the intensity of the ∼110 ppm resonance was observed. Using the fully quantitative relative ratios of areas under the Si versus SiO2 resonances, the following average silicon molar contents have been extracted from spectra deconvolutions (see the Supporting Information section, Figures S10 and S11, for justification of signal quantification and deconvolution details); 79:21 for the pristine sample, 72:28 for the 400 °C treated sample, and 56:44 for the 600 °C sample, respectively. On the basis of these ratios, an average of 6.1, 8, and 12 nm amorphous SiO2 shells was estimated for pristine, 400 and 600 °C samples, respectively. It must be noted that NMR is an averaging bulk technique and perfectly spherical particles with 80 nm diameters was assumed. Therefore, it is possible to have an under or overestimation of the modeled thicknesses depending on the mean particle size distribution. Fourier-transform infrared spectroscopy is a powerful technique for characterizing the IR-active species on the surface as well as in bulk. Figure 3 shows the FTIR spectra of the

Figure 3. FTIR spectra of pristine Si nanoparticles and Si nanoparticles treated at different temperatures.

pristine and treated Si nanoparticles measured in the attenuated total reflection mode. The pristine Si nanoparticles showed a broad feature containing peaks from surface Si−O−Si stretching (1250−850 cm−1) and bending vibrations (850− 750 cm−1) in the 1250−700 cm−1 region. The peaks are broad due to the random packing of the tetrahedral unit of Si−O in the amorphous SiO2 layer.20 As the treatment temperature increases, the peaks become narrower and less convoluted, indicating that the packing form of the Si−O tetrahedral unit becomes more ordered.21 Noticeably, for the 600 and 700 °C treated Si nanoparticles, an additional sharp peak at around 1120 cm−1 appeared. This peak is assigned to oxide in bulk Si by Mawhinney and co-workers,20 indicating that at elevated temperatures, oxygen may further diffuse into bulk silicon to form oxide inside bulk silicon. A more ordered packing structure of the SiO2 layer as well as the existence of oxide in bulk silicon could all affect the resistivity and lithium-ion diffusivity of the SiO2 layer and further affect the electrochemical property of the whole particle. 3.2. Stability of Oxidized Silicon Nanoparticles during the Aqueous Slurry-Making Process. The reaction between

Figure 2. (a) TEM image of the nontreated pristine Si nanoparticles (b−d) Oxygen K map of pristine, 400 °C treated and 600 °C treated Si nanoparticles, and (e) 29Si MAS NMR spectra collected for the pristine and treated samples. Spectra were normalized based on the intensity of the ∼80 ppm resonance.

particle size of 80 nm and an original oxide layer thickness of 2 nm, the calculated thickness increase based on the weight gain is 1 and 4.5 nm for the Si nanoparticles treated at 400 and 600 °C, respectively (see details in Table S1). To obtain more accurate oxide layer thickness, the thicknesses of the SiO2 layer of pristine, 400, and 600 °C heated silicon powders were also investigated using NMR, and the 29Si MAS NMR spectra are shown in Figure 2e. For each spectrum, the sharp resonance at ∼80 ppm is due to elemental silicon and the broad resonance at ∼110 ppm is assigned, according to the literature,19 to amorphous SiO2 on the outer 32729

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces Si nanoparticles and water generates H2 gas.13 To test the stability of our silicon samples in aqueous slurries, we monitored the gas formation after the Si nanoparticles were mixed into the slurry at both room temperature and elevated temperature of 75 °C at which the laminates are usually dried. Three samples were chosen for this test: pristine, 400 °C treated, and 600 °C treated Si samples, as they represent different oxide layer thicknesses. As described in Section 2.4, the Si samples were mixed with LiPAA, C45, and 18 MΩ deionized water, using a centrifugal mixer (Thinky Mixer) at 2000 rpm to obtain good slurry property. The slurries were then transferred into pouches, and the pouches were vacuumsealed immediately. The whole process took about 20 min. The pouches were then placed at room temperature and monitored for 2 weeks for possible swelling behavior that could be caused by gas generation. Figure 4a,b shows the photos of the pouches

samples under room temperature. Our laminates arewere usually dried at 75 °C for 4 h; therefore, we further tested the stability of the slurries at the elevated temperature of 75 °C. Figure 4c shows the pouches after being placed in a 75 °C oven for 3 days (photos were taken after the pouches were cooled to room temperature). We observed significant swelling for the pristine Si sample, slight swelling for the 400 °C treated sample, and no swelling for the 600 °C treated sample. To conclude, all of our silicon samples are fairly stable in room-temperature aqueous slurries. There is not significant wet oxidation and corresponding gas generation when the silicon nanoparticles are subjected to lab-scale mixing and kept at room temperature for 2 weeks. At the elevated temperature of 75 °C, pristine Si nanoparticles showed significant gas generation corresponding to wet oxidation. However, heattreated Si samples with even a slightly thicker oxide layer showed significant improvement in the stability in aqueous slurries. Our results suggest two scenarios: (a) For a small lab-scale slurry-making process, a small amount of slurry of several grams is usually produced at a time and the heat dissipation is relatively good. In this case, the temperature rise of the slurries as well as the time for which the slurries stay at elevated temperatures is usually limited. In this case, there may not be a significant amount of wet oxidation to significantly change the overall oxide layer of the silicon nanoparticles. If there is wet oxidation reaction, thicker silicon oxide layer is expected. To confirm previous finding, we used X-ray photoelectron spectroscopy (XPS) to compare the oxide layer thickness for the pristine and 400 °C treated Si nanoparticles before and after the slurry-making/ electrode fabrication process. Figure S1 shows the XPS Si (2p) results for pristine and 400 °C treated Si nanoparticles in the form of powder (before the slurrymaking/electrode fabrication process) as well as electrode (after the slurry-making/electrode fabrication process). The Si (2p) spectra shows two major peaks corresponding to Si in bulk Si core (denoted as “Si−Si”) and Si in the SiO2 shell (denoted as “SiO2”). For both the powder and electrode made with pristine Si nanoparticles, a much stronger Si signal corresponding to Si in Si bulk core is observed, indicating an overall thinner oxide layer compared to that of the powder and electrode of the 400 °C treated Si. We also integrated the two peaks and compared the relative ratio of area (SiO2) to area (Si−Si). We did not observe any significant change of the area ratio between the powder form and the electrode form for both sets of samples, confirming that for the small lab-scale slurry-making process, there is no significant wet oxidation process occurring for progressively thickening the oxide layer of the silicon nanoparticles during the slurry-making/electrode fabrication process. (b) Though no wet oxidation was observed for Si particles in lab mixing process, the stability of silicon in water is still a valid concern in real application. A large-scale slurrymaking process usually involves mixing of slurries on the kilogram scale. It usually requires much harsher mixing conditions, such as intensive high-energy ball milling for an extended amount of time. During this process, significant temperature rise may happen to the slurries. The slurries may also stay at higher temperature for

Figure 4. Si slurry pouches (a) right after the vacuum seal, (b) after keeping at room temperature for 2 weeks, and (c) after sitting at 75 °C for 3 days.

immediately after vacuum-sealing and then after keeping at room temperature for 2 weeks. We did not observe any pouch swelling during the 2 week previous reports, significant H2 formation was observed within the first 2 h at room temperature,13 we conclude that in our case we did not observe a significant reaction between water and our silicon 32730

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces

suggesting a lot less formation of Li15Si4 phase under the same rate and cycling voltage window of 10 mV and 1.5 V. Comparing the performance of 500 and 600 °C treated Si nanoparticles with the pristine Si nanoparticles, we observed a decreased specific capacity and Coulombic efficiency. The decreased specific capacity is most likely caused by the decrease of the active Si from the heat treatment processes, and the decreased Coulombic efficiency is likely due to the increased parasitic reactions from the irreversible reactions between SiO2 and lithium ions.22,23 Figure S3 shows the first lithiation voltage profile of the pristine and treated Si samples. At the first lithiation, we observed the overall lower lithiation potential as the treatment temperature rises. Du and co-workers have reported that an inactive matrix can put stress to the active material, resulting in less lithiation at the same potential.24,25 Therefore, we also suspect that this decrease of specific capacity may also come from the stress from the significantly thickened SiO2/lithiated SiO2 layer. On the other hand, the Si nanoparticles after 300 and 400 °C heat treatments showed comparable specific capacity and Coulombic efficiency during the formation cycles. Although the pristine Si nanoparticles started with slightly higher reversible capacity, the heat-treated samples retained their capacity better during the three formation cycles, which resulted in very comparable reversible capacity at the third formation cycle. This better capacity retention at formation cycles suggests that the mechanical integrity of the Si nanoparticles may be improved after the treatment. In conclusion, the initial electrochemical performance comparison of different Si samples shows that although too thick of an oxide layer may decrease the initial Coulombic efficiency and reversible specific capacity of the material, a slight increase of the oxide layer thickness does not harm the initial performance. It was noted by Xun and co-workers that when comparing the initial performance, the irreversible capacity loss per surface area should be taken into consideration, as treatment may change the surface area significantly and affect the surface area available for parasitic reactions.12 Here, we measured the BET surface area of our silicon samples and calculated the irreversible capacity loss per surface area for the first cycle. Figure 6 shows the calculated irreversible capacity loss per surface area for samples treated at different temperatures. The detailed BET and irreversible capacity loss calculation results

much longer time. This can result in a significant wet oxidation reaction between silicon and water. In this case, the stability of Si particles at elevated temperatures should be taken into consideration. 3.3. Optimization of the Oxide Layer Thickness for Optimal Electrochemical Performance. 3.3.1. Initial Electrochemical Performance. To obtain the initial electrochemical performance information of heat-treated Si nanoparticles, we cycled the assembled half cells between 10 mV and 1.5 V at C/10 rate for 3 cycles. Figure 5 compares the

Figure 5. (a) Specific capacity and (b) Coulombic efficiency comparison of initial electrochemical performance of Si nanoparticles after different treatments.

electrochemical performance of the initial formation cycles for the Si electrodes with Si nanoparticles of different treatments. Si nanoparticles treated at 700 °C did not show any reversible capacity from silicon and are therefore not shown in Figure 5. The voltage profiles of formation cycles are shown in Figure S2. The first lithiation voltage profiles of pristine and treated Si samples showed a big lithiation overpotential, which corresponds to the lithiation of crystalline Si. The subsequent lithiation profiles for the second and third cycles showed a slopelike feature, which corresponds to the lithiation of amorphous Si. For the pristine, 300 °C treated, and 400 °C treated samples, we observed a flat delithiation profile at 0.44 V, which is characteristic of the formation of Li15Si4 crystalline phase. For the 500 °C treated and 600 °C treated Si samples, the delithiation profiles showed a more slopelike feature,

Figure 6. Irreversible capacity loss per surface area for Si nanoparticles treated at different temperatures. For the 600 °C treated sample, formation data at the C/100 rate was used due to insufficient formation at the C/10 rate. 32731

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces can be found in Table S2. BET surface area measurements show that our oxidation treatment did not result in any significant decrease of the surface area. The calculated irreversible capacity loss per surface area shows that treating the silicon nanoparticles at 300 and 400 °C resulted in slightly decreased irreversible capacity loss per surface area, whereas treating the nanoparticles at 500 and 600 °C increased the irreversible capacity loss per surface area. For 600 °C treated Si, the result from initial formation cycles done at C/100 is shown because we observed an interesting capacity increase over cycles when the electrodes were cycled at C/10, indicating that the electrode was not fully activated at the first three cycles at the C/10 rate. The first cycle capacity loss per surface area at C/10 for 600 °C treated Si samples is shown in Table S2. Irreversible capacity loss is related to the parasitic reactions which could be caused by different factors. For large micronsized particles, particle pulverization and fracture could be a big factor. However, this problem can be largely mitigated when nanoparticles are utilized. Ma and co-workers reported 90 nm as the critical size for particle fracture,26 whereas Liu and coworkers reported that when particle size is smaller than 150 nm, particle fracture is minimized.27 In our case, because most of the particles are in the 100 nm range, the difference in irreversible capacity loss is likely mainly caused by factors such as difference in solid electrolyte interphase (SEI) formation and the parasitic reactions between the oxide layer and the lithium ions. The native oxide layer of silicon is known to contain many dangling bonds due to random packing of Si−O tetrahedral units.28 As the Si nanoparticles get treated in elevated temperatures in air, one may expect the decrease of such dangling bonds and possibly other defects. The decrease of irreversible capacity loss when treated at 300 and 400 °C suggests that the parasitic reactions related to surface defects may be minimized during this heat treatment. The increase of the irreversible capacity loss when treated at 500 and 600 °C is likely caused by the increase of reactions between the significantly increased SiO2 layer and the lithium ions to form lithium silicates.22,23 3.3.2. Rate Performance Comparison for Pristine and 400 °C Treated Si Samples. Because 400 °C treated Si nanoparticles showed significantly improved stability in aqueous slurries as well as uncompromised initial electrochemical performance, for rate and cycle performance tests, we chose 400 °C treated Si nanoparticles for further comparison with the pristine Si nanoparticles. For the rate performance, we compared the delithiation and lithiation rate performances of the pristine and the 400 °C treated Si nanoparticles, which correspond to the discharge and charge rate performances in full cells. For the delithiation rate performance test, the Si/Li half cells were tested at a constant lithiation rate of C/5 and different delithiation rates of C/5, C/ 3, C/2, 1C, and 2C for three cycles each. In the lithiation rate performance test, the delithiation rate was set constant at C/5, whereas the lithiation rate was changed from C/5, C/3, C/2, 1C, to 2C for 3 cycles each. Figure 7a,b shows, respectively, how the capacity is retained when the delithiation rate and lithiation rate is changed. The capacity in relative percentage compared to the capacity obtained at the C/5 rate (second cycle) was used. Results in absolute specific capacity are shown in Figure S4. A significant improvement of rate performance was observed for the 400 °C treated silicon samples in the delithiation rate performance test, whereas the lithiation rate test showed a much similar performance, with a slight

Figure 7. (a) Delithiation and (b) lithiation rate performance of pristine and 400 °C treated Si nanoparticles.

improvement observed for 400 °C treated silicon nanoparticles. Meanwhile, we also observed that for both pristine and 400 °C treated silicon nanoparticles, the delithiation rate test showed better capacity retention at high rates than the lithiation rate test. This asymmetric rate behavior of Si anodes (better rate performance at delithiation than lithiation) has been reported by Li and co-workers and is attributed to the potentialconcentration profile associated with the active material and the Ohmic voltage shift under high currents.29 Compared to other materials, silicon does show significant amount of capacity decrease from cycle to cycle due to structure degradation.1 To confirm that the different rate performances for the pristine and 400 °C treated Si samples is indeed caused by the rate and not just degradation of the electrodes, we tested another set of cells directly to the 2C rate after three C/10 formation cycles and one C/5 cycle. The cells were then cycled back to the C/5 rate for information on how much specific capacity can still be recovered. Figure S5 shows the result of this test. For the cells with the pristine Si nanoparticles, about 75% of C/5 capacity was achieved at the 2C delithiation rate, whereas for the cells with 400 °C treated Si, close to 95% of the C/5 capacity was achieved at the 2C delithiation rate. When the cells were cycled back to the C/5 rate, most of the capacity was recovered, with some loss caused by structure degradation during cycling. This experiment confirms that the rate performance difference is not due to the degradation of the electrodes but indeed a real reflection of the material’s ability to discharge at high rates. 32732

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces

It should also be noted that when the delithiation rate was being changed and the lithiation rate was set constant, the capacity of the electrodes underwent constant fading over cycles, whereas when the lithiation rate was being changed and the delithiation rate was set constant, the capacity of the electrodes was fairly stable for the three cycles tested at the same lithiation rate. This suggests that the silicon nanoparticles are subject to more mechanical integrity challenges during delithiation than lithiation. 3.3.3. Cycle Performance Comparison for Pristine and 400 °C Treated Si Samples. We further compared the cycle performance of the 400 °C treated Si nanoparticles with the pristine silicon nanoparticles in a full-cell configuration, with LiNi0.5Co0.2Mn0.3O2 (NCM523) used as the cathode. To get full-cell cycle performance information, we cycled the assembled full cells at the C/3 rate for 50 cycles after three formation cycles at the C/10 rate. Area specific impedance (ASI) was measured every 10 cycles during the cycle test, and the initial ASI after formation cycles are shown in Figure S7. The area specific impedance of both sets of cells are comparable. Table 1 shows the capacity retention for different electrodes cycled at different voltage windows.

To understand the rate performance improvement, we consider the mechanical property of the Si nanoparticles with oxide layers of different thicknesses. The SiO2 layer around Si has been reported to reduce the volume expansion of Si by Sim and co-workers,30 demonstrating that extra stress can be provided by such a layer during the cycling of Si particles. In our case, we propose that this extra stress provided by the lithiated SiO2 layer may also facilitate the lithium diffusion. An in situ Raman study by Zeng and co-workers has shown that when Si nanoparticle with a thin oxide layer undergoes lithiation, the SiO2 layer is lithiated first, causing tensile stress of ∼0.2 GPa to the Si core.31 Because Li diffusivity has been demonstrated to be dependent on the lattice rigidity, with less lattice rigidity leading to faster lithium diffusion,32,33 we propose that when the extra tensile stress by the thicker lithiated oxide layer is present, the lattice is less rigid and lithium diffusion becomes faster and thus the improved rate performance. During the lithiation rate test, a much smaller difference between the 400 °C treated Si and pristine Si electrodes was observed. This suggests that the tensile stress does not play a major role during lithiation. During lithiation, compressive stress is caused by the lithiation process,27,31 which leads to harder lithium diffusion into the Si core. This compressive stress can then be mitigated by deformation of the particles.27 During lithiation, compressive stress caused by lithiation of Si likely plays a much larger role than the tensile stress from the lithiated SiO2 layer. The improved rate performance suggests that the slightly thicker oxide layer at the 400 °C treated silicon surface facilitated the delithiation and lithiation processes. Because bulk silica is electronically insulating, this rate improvement may seem counterintuitive. Bulk silica is insulating; however, for a very thin silica layer, its electronic resistance may be significantly decreased due to its thickness and less-ordered structure. We performed alternating current impedance analysis for 50% state-of-charge Si half cells after formation cycles, and the results are shown in Figure S6. The results show that compared with electrodes with pristine Si nanoparticles, electrodes with 400 °C treated Si nanoparticles showed comparable impedance at high frequency, suggesting overall comparable electronic conductivity. The overall slightly smaller semicircles also suggest that the lithium diffusion is better in the Si electrodes with 400 °C treated Si nanoparticles, in agreement with our previous hypothesis. When considering the effect of the thin silica layer on the rate performance, one should consider the concurrent effects of electronic conduction and ionic transport, as suggested by Zhang and co-workers.34 The silica thin film with a larger thickness of 9 nm has also been used as a solid electrolyte for lithium-ion transportation between the electrodes for all solid-state Li-ion batteries, demonstrating its ability to transport lithium ions.35 In fact, calculations on the lithium diffusivity for silica and lithiated silica show that the Li diffusivity ranges from 10−8 m2/s for silica to 10−11 m2/s for lithiated silica,34 whereas the measured and calculated Li diffusivity in lithiated silicon is in the range of 10−15−10−17 m2/s.29,36,37 This Li diffusivity is much higher in silica and lithiated silica than in silicon and lithiated silicon, suggesting that Li diffusivity in silica should not be the ratelimiting step during lithiation and delithiation of Si nanoparticles. Therefore, a slightly thicker oxide layer does not necessarily affect the electronic conduction and ionic transport of the Si nanoparticles.

Table 1. Capacity Retention of Full Cells Cycled at Different Voltage Windows capacity retention

2.5−4.3 V

3.2−4.3 V

Si, pristine Si, 400 °C treated

52.91 ± 0.12% 59.70 ± 0.94%

72.23 ± 0.35% 78.31 ± 0.23%

Figure 8a,b shows the cycle performance of the 400 °C treated and the pristine silicon nanoparticles between 2.5 and 4.3 V. The anode capacity is used here because the performance of anodes is compared in this study. The specific anode capacity in full-cell tests appears lower than the specific capacity in halfcell tests. This is because in full-cell tests, an N/P ratio of 1.1− 1.2 is utilized so that the anode has some extra capacity and lithium plating on the anode can be prevented. The presence of extra silicon in the anode leads to the lower specific anode capacity. Compared to pristine silicon nanoparticles, the 400 °C treated Si nanoparticles showed an improvement of capacity retention from 52.91% (3 cell average) to 59.70% (3 cell average) when cycled between 2.5 V and 4.3 V. It should also be noted that a significant improvement in capacity retention was observed for both sets of cells when we raised the lower cutoff voltage from 2.5 to 3.2 V, as shown in Figure 8c,d. In this case, about 17% less lithium was cycled compared to the case where the cells were cycled between 2.5 and 4.3 V due to the smaller cycling voltage window. Compared with the pristine silicon cells, the 400 °C treated Si full cells showed an improvement of capacity retention from 72.23% (3 cell average) to 78.31% (2 cell average) when cycled between 3.2 V and 4.3 V. At the end of the 50 cycles, the cells cycled with the 3.2 V lower cutoff voltage retain a larger specific capacity than the cells cycled with a 2.5 V lower cutoff voltage. The much better capacity retention with a 3.2 V lower cutoff voltage can be explained by the higher Coulombic efficiency observed over cycling, as shown in Figure 8b,d. This higher Coulombic efficiency can be caused by the better structural stability when less lithium is cycled. Another likely reason for the higher Coulombic efficiency is the better electrochemical stability of the solid electrolyte interphase layer of silicon. When the 2.5 V lower cutoff voltage is utilized, the solid electrolyte interphase 32733

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces

Figure 8. (a, b) Capacity retention and Coulombic efficiency of pristine and 400 °C treated Si nanoparticles cycled between 2.5 and 4.3 V (c, d) capacity retention and Coulombic efficiency of pristine and 400 °C treated Si nanoparticles cycled between 3.2 and 4.3 V.

S8 shows the half-cell cycle performance for pristine, 300, 400, 500, and 600 °C treated Si electrodes for 40 cycles. For 500 and 600 °C treated Si electrodes, although the reversible capacity is compromised in both cases in the initial formation cycles, both treated samples showed improved capacity retention during the 40 cycles of the cycle test. This further supports that a thicker oxide layer improves the mechanical stability of the particles during cycling thus improving the capacity retention. Note that for pristine and 400 °C treated Si nanoparticles, better capacity retention is observed in half cells compared with full-cell testing results. In half-cell tests, the lithium inventory is unlimited; therefore, the effect of irreversible lithium consumption caused by parasitic reactions (such as new SEI formation) on capacity fading is not properly reflected. Full-cell testing results better reflect the capacity retention in practical applications.

of silicon is more likely to be subject to a more oxidizing potential and therefore is more likely to be oxidized. This oxidation and subsequent reformation of the SEI layer at each cycle leads to extra irreversible lithium consumption which leads to poor capacity retention. The Coulombic efficiency comparison in Figure 8b,d shows that the initial Coulombic efficiency of the 400 °C treated Si full cells is on an average higher than the Coulombic efficiency of the pristine Si full cell for both voltage windows. However, when cycled between 2.5 and 4.3 V, the Coulombic efficiency decreased for 400 °C treated Si full cells before it rose again as the cells cycled. This phenomenon was not observed for the 400 °C treated Si full cells cycled between 3.2 and 4.3 V. This suggests that the lower cutoff voltage may degrade the oxide layer in some way. More study needs to be conducted to fully understand this phenomenon. The improved cycle performance that we observed agrees with the previous work conducted by Sim and co-workers, where they studied the impact of SiO2 coating thickness on the capacity retention of 1.4 μm Si particles and found that with a thicker SiO2 layer, the capacity retention is improved.30 They attributed the improvement to the suppression of the volume expansion by the SiO2 coating to the particles. In our case, the capacity retention improvement may also be attributed to the improved mechanical integrity of the particles, as better mechanical property makes particles less likely to crack during the cycling; thus, less lithium consumption is caused by SEI formation. In fact, we also observed improved capacity retention with 500 and 600 °C treated Si nanoparticles. Figure

4. CONCLUSIONS In this work, we investigated the stability of partially oxidized silicon nanoparticles in aqueous slurries and optimized the oxide layer thickness of silicon nanoparticles for optimal electrochemical performance. The oxidized Si nanoparticles demonstrated improved stability in aqueous slurries at elevated temperatures. We also found that with a slightly increased oxide layer thickness, the initial electrochemical performance was not compromised. Furthermore, an improvement in the rate performance and full-cell cycle performance can also be achieved. 32734

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces



(5) Wu, H.; Chan, G.; Choi, J. W.; Ryu, I.; Yao, Y.; McDowell, M. T.; Lee, S. W.; Jackson, A.; Yang, Y.; Hu, L. B.; Cui, Y. Stable Cycling of Double-Walled Silicon Nanotube Battery Anodes through SolidElectrolyte Interphase Control. Nat. Nanotechnol. 2012, 7, 310−314. (6) Magasinski, A.; Dixon, P.; Hertzberg, B.; Kvit, A.; Ayala, J.; Yushin, G. High-Performance Lithium-Ion Anodes Using a Hierarchical Bottom-up Approach. Nat. Mater. 2010, 9, 353−358. (7) Liu, N.; Wu, H.; McDowell, M. T.; Yao, Y.; Wang, C. M.; Cui, Y. A Yolk-Shell Design for Stabilized and Scalable Li-Ion Battery Alloy Anodes. Nano Lett. 2012, 12, 3315−3321. (8) Park, C. M.; Kim, J. H.; Kim, H.; Sohn, H. J. Li-Alloy Based Anode Materials for Li Secondary Batteries. Chem. Soc. Rev. 2010, 39, 3115−3141. (9) Raider, S. I.; Flitsch, R.; Palmer, M. J. Oxide-Growth on Etched Silicon in Air at Room-Temperature. J. Electrochem. Soc. 1975, 122, 413−418. (10) Morita, M.; Ohmi, T.; Hasegawa, E.; Kawakami, M.; Ohwada, M. Growth of Native Oxide on a Silicon Surface. J. Appl. Phys. 1990, 68, 1272−1281. (11) Xun, S.; Song, X.; Grass, M. E.; Roseguo, D. K.; Liu, Z.; Battaglia, V. S.; Liu, G. Improved Initial Performance of Si Nanoparticles by Surface Oxide Reduction for Lithium-Ion Battery Application. Electrochem. Solid-State Lett. 2011, 14, A61−A63. (12) Xun, S.; Song, X.; Wang, L.; Grass, M. E.; Liu, Z.; Battaglia, V. S.; Liu, G. The Effects of Native Oxide Surface Layer on the Electrochemical Performance of Si Nanoparticle-Based Electrodes. J. Electrochem. Soc. 2011, 158, A1260−A1266. (13) Touidjine, A.; Morcrette, M.; Courty, M.; Davoisne, C.; Lejeune, M.; Mariage, N.; Porcher, W.; Larcher, D. Partially Oxidized Silicon Particles for Stable Aqueous Slurries and Practical Large-Scale Making of Si-Based Electrodes. J. Electrochem. Soc. 2015, 162, A1466− A1475. (14) Li, J.; Le, D. B.; Ferguson, P. P.; Dahn, J. R. Lithium Polyacrylate as a Binder for Tin-Cobalt-Carbon Negative Electrodes in Lithium-Ion Batteries. Electrochim. Acta 2010, 55, 2991−2995. (15) Magasinski, A.; Zdyrko, B.; Kovalenko, I.; Hertzberg, B.; Burtovyy, R.; Huebner, C. F.; Fuller, T. F.; Luzinov, I.; Yushin, G. Toward Efficient Binders for Li-Ion Battery Si-Based Anodes: Polyacrylic Acid. ACS Appl. Mater. Interfaces 2010, 2, 3004−3010. (16) Yoshida, S.; Masuo, Y.; Shibata, D.; Haruta, M.; Doi, T.; Inaba, M. Adsorbed Water on Nano-Silicon Powder and Its Effects on Charge and Discharge Characteristics as Anode in Lithium-Ion Batteries. J. Electrochem. Soc. 2017, 164, A6084−A6087. (17) McNeill, I. C.; Sadeghi, S. M. T. Thermal-Stability and Degradation Mechanisms of Poly(Acrylic Acid) and Its Salts.1. Poly(Acrylic Acid). Polym. Degrad. Stab. 1990, 29, 233−246. (18) Trask, S. E.; Li, Y.; Kubal, J. J.; Bettge, M.; Polzin, B. J.; Zhu, Y.; Jansen, A. N.; Abraham, D. P. From Coin Cells to 400 Mah Pouch Cells: Enhancing Performance of High-Capacity Lithium-Ion Cells Via Modifications in Electrode Constitution and Fabrication. J. Power Sources 2014, 259, 233−244. (19) MacKenzie, K.; Smith, M. E. Multinuclear Solid-State Nuclear Magnetic Resonance of Inorganic Materials, 1st ed.; Pergamon: The Netherlands, 2002. (20) Mawhinney, D. B.; Glass, J. A.; Yates, J. T. FTIR Study of the Oxidation of Porous Silicon. J. Phys. Chem. B 1997, 101, 1202−1206. (21) Lippincott, E. R.; Vanvalkenburg, A. V.; Weir, C. E.; Elmer, N. Bunting Infrared Studies on Polymorphs of Silicon Dioxide and Germanium Dioxide. J. Res. Natl. Bur. Stand. 1958, 61, 61−70. (22) Philippe, B.; Dedryvere, R.; Gorgoi, M.; Rensmo, H.; Gonbeau, D.; Edstrom, K. Role of the LiPF6 Salt for the Long-Term Stability of Silicon Electrodes in Li-Ion Batteries - a Photoelectron Spectroscopy Study. Chem. Mater. 2013, 25, 394−404. (23) Philippe, B.; Dedryvere, R.; Gorgoi, M.; Rensmo, H.; Gonbeau, D.; Edstrom, K. Improved Performances of Nanosilicon Electrodes Using the Salt LiFSI: A Photoelectron Spectroscopy Study. J. Am. Chem. Soc. 2013, 135, 9829−9842.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsami.7b09149. Estimation of oxide layer thickness on the basis of TEM and weight gain, X-ray photoelectron spectroscopy (XPS) measurements for pristine and 400 °C treated Si samples in the forms of powder and electrode, voltage profiles of the initial formation cycles for pristine silicon and silicon samples treated at different temperatures, irreversible capacity loss per surface area of the first formation cycle for all Si samples, first lithiation voltage profiles for different Si samples, rate test for pristine and 400 °C treated Si nanoparticles in absolute specific capacity, additional rate test for pristine and 400 °C treated Si samples, Nyquist plots for half cells with pristine and 400 °C treated Si nanoparticles, area specific impedance for full cells with pristine and 400 °C treated Si nanoparticles, half-cell cycle performance for different Si samples, 29Si magic angle spinning (MAS) NMR experiments and data analysis (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Wenquan Lu: 0000-0001-8655-8256 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors gratefully acknowledge support from the U.S. Department of Energy (DOE), Vehicle Technologies Office. Argonne National Laboratory is operated for DOE Office of Science by UChicago Argonne, LLC, under contract number DE-AC02-06CH11357. Support from the cell analysis, modeling, and prototyping (CAMP) Facility at Argonne National Laboratory is gratefully acknowledged. The TEM work was carried out at Center for Nanomaterials at Argonne National Laboratory. This work also made use of instruments in the Electron Microscopy Service at Research Resources Center of the University of Illinois at Chicago.



REFERENCES

(1) Obrovac, M. N.; Chevrier, V. L. Alloy Negative Electrodes for LiIon Batteries. Chem. Rev. 2014, 114, 11444−11502. (2) Xu, Q.; Li, J. Y.; Sun, J. K.; Yin, Y. X.; Wan, L. J.; Guo, Y. G. Watermelon-Inspired Si/C Microspheres with Hierarchical Buffer Structures for Densely Compacted Lithium-Ion Battery Anodes. Adv. Energy Mater. 2017, 7, No. 1601481. (3) Zhou, X.; Yin, Y. X.; Wan, L. J.; Guo, Y. G. Facile Synthesis of Silicon Nanoparticles Inserted into Graphene Sheets as Improved Anode Materials for Lithium-Ion Batteries. Chem. Commun. 2012, 48, 2198−2200. (4) Zhou, X. S.; Yin, Y. X.; Wan, L. J.; Guo, Y. G. Self-Assembled Nanocomposite of Silicon Nanoparticles Encapsulated in Graphene through Electrostatic Attraction for Lithium-Ion Batteries. Adv. Energy Mater. 2012, 2, 1086−1090. 32735

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736

Research Article

ACS Applied Materials & Interfaces (24) Du, Z. J.; Dunlap, R. A.; Obrovac, M. N. Structural and Electrochemical Investigation of FexSi1‑X Thin Films in Li Cells. J. Electrochem. Soc. 2016, 163, A2011−A2016. (25) Du, Z. J.; Hatchard, T. D.; Dunlap, R. A.; Obrovac, M. N. Combinatorial Investigations of Ni-Si Negative Electrode Materials for Li-Ion Batteries. J. Electrochem. Soc. 2015, 162, A1858−A1863. (26) Ma, Z. S.; Li, T. T.; Huang, Y. L.; Liu, J.; Zhou, Y. C.; Xue, D. F. Critical Silicon-Anode Size for Averting Lithiation-Induced Mechanical Failure of Lithium-Ion Batteries. RSC Adv. 2013, 3, 7398−7402. (27) Liu, X. H.; Zhong, L.; Huang, S.; Mao, S. X.; Zhu, T.; Huang, J. Y. Size-Dependent Fracture of Silicon Nanoparticles During Lithiation. ACS Nano 2012, 6, 1522−1531. (28) Helms, C. R.; Poindexter, E. H. The Silicon Silicon-Dioxide System - Its Microstructure and Imperfections. Rep. Prog. Phys. 1994, 57, 791−852. (29) Li, J. C.; Dudney, N. J.; Xiao, X. C.; Cheng, Y. T.; Liang, C. D.; Verbrugge, M. W. Asymmetric Rate Behavior of Si Anodes for Lithium-Ion Batteries: Ultrafast De-Lithiation Versus Sluggish Lithiation at High Current Densities. Adv. Energy Mater. 2015, 5, No. 1401627. (30) Sim, S.; Oh, P.; Park, S.; Cho, J. Critical Thickness of SiO2 Coating Layer on Core@Shell Bulk@Nanowire Si Anode Materials for Li-Ion Batteries. Adv. Mater. 2013, 25, 4498−4503. (31) Zeng, Z. D.; Liu, N. A.; Zeng, Q. S.; Lee, S. W.; Mao, W. L.; Cui, Y. In Situ Measurement of Lithiation-Induced Stress in Silicon Nanoparticles Using Micro-Raman Spectroscopy. Nano Energy 2016, 22, 105−110. (32) Chou, C. Y.; Hwang, G. S. On the Origin of the Significant Difference in Lithiation Behavior between Silicon and Germanium. J. Power Sources 2014, 263, 252−258. (33) Ning, F. H.; Li, S.; Xu, B.; Ouyang, C. Y. Strain Tuned Li Diffusion in LiCoO2 Material for Li Ion Batteries: A First Principles Study. Solid State Ionics 2014, 263, 46−48. (34) Zhang, Y.; Li, Y. J.; Wang, Z. Y.; Zhao, K. J. Lithiation of SiO2 in Li-Ion Batteries: In Situ Transmission Electron Microscopy Experiments and Theoretical Studies. Nano Lett. 2014, 14, 7161−7170. (35) Ariel, N.; Ceder, G.; Sadoway, D. R.; Fitzgerald, E. A. Electrochemically Controlled Transport of Lithium through Ultrathin SiO2. J. Appl. Phys. 2005, 98, No. 023516. (36) Wang, M.; Xiao, X. R.; Huang, X. S. Study of Lithium Diffusivity in Amorphous Silicon Via Finite Element Analysis. J. Power Sources 2016, 307, 77−85. (37) Pell, E. M. Diffusion Rate of Li in Si at Low Temperatures. Phys. Rev. 1960, 119, 1222−1225.

32736

DOI: 10.1021/acsami.7b09149 ACS Appl. Mater. Interfaces 2017, 9, 32727−32736