GB2512230B

GB2512230B - Energy storage devices

Info

Publication number: GB2512230B
Application number: GB1411191.8A
Authority: GB
Inventors: Rojeski Ronald; Klankowski Steven; Li Jun
Original assignee: Individual
Current assignee: Individual
Priority date: 2011-12-21
Filing date: 2012-12-21
Publication date: 2019-08-28
Anticipated expiration: 2032-12-21
Also published as: GB201411191D0; KR20140128298A; WO2013096931A1; CN104145355B; CN106910869A; KR102036196B1; GB2512230A; CN104145355A

Description

Energy Storage Devices

CROSS-REFERENCE TO RELATED APPLICATIONS

[001] This application claims benefit and priority to U.S. provisional patent applications: 61/578,545 filed Dec. 21, 2011, 61/603,833 filed Feb. 27, 2012, 61/615,179 filed Mar. 23, 2012, 61/667,876 filed Jul. 3, 2012, and 61/677,317 filed Jul. 30, 2012.

[002] The disclosures of all the above provisional and non-provisional patent applications are hereby incorporated herein by reference.

BACKGROUND

[003] Field of the Invention [004] The invention is in the field of energy storage devices, including but not limited to batteries, capacitors and fuel cells.

[005] Related Art [006] Rechargeable lithium ion batteries are key electrical energy storage devices for power supply in portable electronics, power tools, and future electric vehicles. Improving the specific energy capacity, charging/discharging speed, and cycling lifetime is critical for their broader applications.

[007] In current commercial Li-ion batteries, graphite or other carbonaceous materials are used as the anodes which have a theoretical capacity limit at 372 mAh/g by forming fully intercalated LiCg compound. In contrast, silicon has a much higher theoretical specific capacity of 4,200 mAh/g by forming fully lithiated alloy Li4.4Si. However, the large volume expansion of lithiated Si by up to -300% causes great structural stress that in the past inevitably lead to fractures and mechanical failure, which significantly limited the lifetime of Si anodes.

SUMMARY

[0008] A power storage device includes a hybrid core-shell NW (nano-wire) architecture in a high-performance Li-ion anode by incorporating an array of vertically aligned carbon nanofibers (VACNFs) coaxially coated with a layer of amorphous silicon. The vertically aligned CNFs include multiwalled carbon nanotubes (MWCNTs), which are optionally grown on a Cu substrate using a DC-biased plasma chemical vapor deposition (PECVD) process. The carbon nanofibers (CNFs) grown by this method can have a unique interior morphology distinguishing them from the hollow structure of common MWCNTs and conventional solid carbon nanofibers. One of the distinguishing characteristics is that these CNFs optionally consist of a series of bamboo-like nodes across the mostly hollow central channel. This microstructure can be attributed to a stack of conical graphitic cups discussed further elsewhere herein. In larger length scale, these PECVD-grown CNFs are typically uniformly aligned normal to the substrate surface and are well separated from each other. They may be without any entanglement or with minimal entanglement, and thus form a brush-like structure referred to as a VACNF array. The diameter of individual CNFs can be selected to provide desired mechanical strength so that the VACNF array is robust and can retain its integrity through Si deposition and wet electrochemical tests.

[0009] The invention provides an energy storage system comprising a conductive substrate; a plurality of aligned carbon nanofibers grown on the substrate with a layer of intercalation material disposed on the carbon nanofibers, the carbon nanofibers each including a plurality of multi-walled carbon nanotubes, wherein the carbon nanotubes are disposed such that ion intercalation can occur between walls of the nanotubes through the side walls of the carbon nanofibers, with the carbon nanofibers including a stacked-cone structure; and an electrolyte including one or more charge carriers. In some embodiments, the layer of intercalation material has a feather-like structure resulting from the stacked-cone structure. In some embodiments, the layer of intercalation material comprises feather-like structures of silicon filled with surface electrolyte interphase. In some embodiments, the intercalation material has a nominal thickness of between 0.01 and 25 μπι. In some embodiments, the layer of intercalation material includes nano fiber/intercalation material complexes, some of the nanofiber/ intercalation material complexes including one nanofiber and some of the nanofiber/ intercalation material complexes including two nanofibers. In some embodiments, the layer of intercalation material on the carbon nanofibers comprises silicon, the silicon having a nominal thickness of between 0.01 pm and 25 pm. In some embodiments, the carbon nanofibers are between 3.0 and 200 pm in length. In some embodiments, the intercalation material includes silicon.

BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIGs. IA and IB illustrate a CNF array comprising a plurality of CNF grown on a substrate, according to various embodiments of the invention.

[0014] FIGs. 2A-2C illustrate a plurality of vertically aligned CNFs in different states, according to various embodiments of the invention.

[0015] FIGs. 3A-3C illustrate details of a CNF, according to various embodiments of the invention. [0016] FIG. 4 illustrates a schematic of the stacked-cone structure of a CNF, according to the invention.

[0017] FIGs. 5A-5C illustrate an electrochemical characterization of ~3 pm long CNFs, according to various embodiments of the invention.

[0018] FIGs. 6A-6C illustrates scanning electron microscopy images of 3 pm long CNFs, according to various embodiments of the invention.

[0019] FIGs. 7A-7C illustrate results obtained using CNFs including a Si layer as Li-ion battery anodes, according to various embodiments of the invention.

[0020] FIG. 8 illustrates how the capacity of a CNF array varies with charging rate, according to various embodiment of the invention.

[0021] FIG. 9A illustrates Raman spectra of CNF arrays, according to various embodiments of the invention.

[0022] FIGs. 10A-10C shows the variation of Li+ insertion-extraction capacities and the coulombic efficiency over 15 charge-discharge cycles, according to various embodiments of the invention.

[0023] FIGs. 11 A-l 1C show scanning electron microscopy images of freshly prepared CNF arrays, according to various embodiments of the invention.

[0024] FIG. 1 ID shows a cross-section of a nanofiber/silicon complex including more than one CNF. [0025] FIG. 12 illustrates a carbon nano-fiber array including fibers of 10 um in length, according to various embodiments of the invention.

[0026] FIG. 13 illustrates methods of producing CNF arrays, according to various embodiments of the invention.

DETAILED DESCRIPTION

[0027] FIGs. IA and IB illustrate a CNF Array 100 comprising a plurality of CNF 110 grown on a conductive Substrate 105, according to various embodiments of the invention. In FIG. IA the CNF Array 100 is shown in the Li extracted (discharged) state and in FIG. IB the CNF Array 100 is shown in the Li inserted (charged) state. The CNF 110 in these and other embodiments discussed herein are optionally vertically aligned. The CNF 110 are grown on a

Substrate 105 of Cu using a DC-biased plasma chemical vapor deposition (PECVD) process. As discussed above, the CNFs 110 grown by this method can have a unique morphology that includes a stack of conical graphitic structures similar to stacked cups or cones or a spiral. This creates a very fine structure that facilitates lithium intercalation. This structure is referred to here as the “stacked-cone” structure elsewhere herein. In larger length scale, these CNFs 110 are typically uniformly aligned normal to the substrate surface and are well separated from each other. The diameter of individual CNFs can be selected to provide desired mechanical strength so that the CNF Array 100 is robust and can retain its integrity through Si deposition and wet electrochemical cycles. A seed layer is optionally employed for growing CNFs 110 on Substrate 105. In use the CNF Array 100 is placed in contact with an Electrolyte 125 including one or more charge carriers, such as a lithium ion. The CNFs 110 are configured such that some of Electrolyte 125 is disposed between CNFs 110 and/or can ready Substrate 105 via gaps between CNFs 110.

[0028] The diameter of individual CNFs 110 illustrated in FIGs. IA and IB are nominally between 100 and 200 nm, although diameters between 75 and 300 nm, or other ranges are possible. CNFs 110 are optionally tapered along their length. The CNFs 110 produced using the techniques discussed herein have excellent electrical conductivity (σ = ~2.5xl05 S/m) along the axis and do form firm Ohmic contact with Substrate 105. The open space between the CNFs 110 enables a Silicon Layer 115 to be deposited onto each CNFs to form a gradually thinned coaxial shell with a mass at a Tip 120 of the CNF 110. This design enables the whole Silicon Layer 115 to be electrically connected through the CNF 110 and to remain fully active during charge-discharge cycling. The expansion that occurs on alloying of lithium with Silicon Layer 115 can be easily accommodated in the radial direction, e.g. perpendicular to the long dimension of the CNFs 110. The charge and discharge capacity and cycling stability of non-Si-coated CNFs 110 and Si-coated CNFs 110 can be compared. The addition of Silicon Layer 115 provided a remarkable LT insertion (charge) capacity up to 3938 mAh/gst at the C/2 rate and retained 1944 mAh/gsi after 110 cylces. This charge/discharge rate and the corresponding capacity are significantly higher than previous architectures using Si nanowires or hybrid Si-C nanostructures. FIGs. IA and IB are perspective views.

[0029] In various embodiments, from 0.01 up to 0.5, 1.0, 1.5, 2.5, 3.0, 4.0, 10, 20, 25 pm (or more) nominal Si thickness can be deposited onto 3 pm long CNFs 110 to form CNF Arrays 100 such as those illustrated in FIG.s IA and IB. Likewise, in various embodiments, from 0.01 up 0.5, 1.0, 1.5, 2.5, 3.0, 4.0, 10, 20, 25 pm (or more) nominal Si thickness can be deposited onto 10 pm long CNFs 110 to form CNF Arrays 100. In some embodiments, the nominal thickness of Si is between 0.01 pm and the mean distance between CNFs 110.

[0030] Using CNF Arrays 100, Li ion storage with up to -4,000 mAh/g mass-specific capacity at C/2 rate is achieved. This capacity is significantly higher than those obtained with Si nanowires alone or other Si-nanostructured carbon hybrids at the same power rate. The improved performance is attributed to the fully activated Si shell due to effective charge collection by CNFs 110 and short Li path length in this hybrid architecture. Good cycling stability has been demonstrated in over 110 cycles. In various embodiments the storage capacity of Li ion storage of CNF Arrays 100 is approximately 750, 1500, 2000, 2500, 3000, 3500 or 4000 mAh per gram of Si, or within any range between these values. As used herein, the term “nominal thickness” (of e.g., Si) is the amount of Si that would produce a flat layer of Si, of the said thickness, on Substrate 105. For example, a nominal thickness of Si of 1.0 pm is an amount of Si that would result in a 1.0 pm thick layer of Si if deposited directly on Substrate 105. Nominal thickness is reported because it can easily be measured by weight using methods know in the art. A nominal thickness of 1.0 pm will result in a smaller thickness of Si Layer 115 on CNFs 110 because the Si is distributed over the greater area of the CNFs 110 surfaces.

[0031] FIGs. 2A-2C illustrate CNF Array 100 having an average fiber length of approximately 3 pm, according to various embodiments of the invention. FIGs. 2A-2C are scanning electron microscopy (SEM) images. FIG. 2A shows a plurality of vertically aligned CNFs 110 without Silicon Layer 115. FIG. 2B shows a plurality of vertically aligned CNFs 110 including Silicon Layer 115. FIG. 2C shows a plurality of vertically aligned CNFs 110 in the extracted (discharged) state after experiencing 100 lithium charge-discharge cycles. The CNFs 110 are firmly attached to a Cu Substrate 105 with essentially uniform vertical alignment and a random distribution on the surface of the substrate. The samples used in this study have an average areal density of 1.1 lxl09 CNFs/cm2 (counted from top-view SEM images), corresponding to an average nearest-neighbor distance of -330 nm. The average length of the CNFs 110 in Figure 2 is ~3.0 pm with >90% of CNFs in the range of 2.5 to 3.5 pm in length. The diameter spreads from ~80 nm to 240 nm with an average of ~ 147 nm. An inverse teardrop shaped Ni catalyst at Tip 120 presents at the tip of each CNF 110 capping the hollow channel at the center of the CNF, which promoted the tip growth of CNF 110 during the PECVD process. The size of the Ni catalyst nanoparticles defined the diameter of each CNFs 110. Longer CNFs 110, up to 10 pm, were also employed in some studies to be discussed in later sections.

[0032] In various embodiments the average nearest neighbor distance can vary between 200-450nm, 275-385 nm, 300-360nm, or the like. Further, the average length of the CNFs 110 can be between approximately 2-20, 20-40, 40-60, 60-80, 80-100, 100-120, 120-250 (pm), or more. Standard carbon nanofibers as long as a millimeter long are known in the art. In various embodiments, the average diameter can vary between approximately 50-125, 100-200, 125-175 (nm), or other ranges.

[0033] An amorphous Si Layer 115 was deposited onto the CNF Array 100 by magnetron sputtering. The open structure of brush-like CNF Arrays 100 made it possible for Si to reach deep down into the array and produce conformal structures between the CNFs 110. Asa result, it formed a thick Si coating at the CNF tip followed by a gradually thinned coaxial Si shell around the lower portion of the CNF, presenting an interesting tapered core-shell structure similar to a cotton swab. The amount of Si deposition is characterized by the nominal thickness of Si films on a flat surface using a quartz crystal microbalance (QCM) during sputtering. The Li insertion/extraction capacities were normalized to the total Si mass derived from the nominal thickness. At 0.50 pm nominal thickness, the Si-coated CNFs 110 were well-separated from each other, forming an open core-shell CNF array structure (shown in FIG. 2B). This structure allowed electrolyte to freely accessing the entire surface of the Si Layer 115. In the embodiment illustrated the average tip diameter was -457 nm in comparison with the -147 nm average diameter of the CNFs 110 prior to application of the Si Layer 115. The average radial Si thickness at the Tip 120 was estimated to be -155 nm. This was apparently much smaller than the 0.50 pm nominal Si thickness since most Si spread along the full length of CNFs. Other radial Si thicknesses in the range of 10-1000, 20-500, 50-250, 100-200 (nm) or different ranges are found in alternative embodiments. As discussed elsewhere herein, the stacked-cone of CNFs 110 provides additional fine structure to the Si Layer 115. The stacked-cone structure is optionally the result of a spiral growth pattern that produces the stacked-cone structure when viewed in cross-section.

[0034] The transmission electron microscopy (TEM) images in FIGs. 3A - 3C further illustrate the structural details of Si-coated CNFs 110. A Si Layer 115 of -390 nm Si was produced directly above the Tip 120 of a -210 nm diameter CNF 110. The largest portion of the cotton-swab-shaped Si Layer 115 was -430 nm in diameter which appeared near the very end of the Tip 120. The coaxial Si Layer 115 around the CNF 110 showed a feather-like texture with modulated contrast, clearly different from the uniform Si deposits above the tip (see FIG. 3A). This is likely a result of the stacked-cone microstructure of the PECVD-grown CNFs 110. It is known from the literature that such CNFs 110 include unevenly stacked cuplike graphitic structures along the CNF 110 center axis. The use of such variations in the diameter of CNFs 110 was previously disclosed in commonly owned U.S. Patent Application Ser. No. 12/904,113 filed Oct. 13. 2010. The stacked-cone structure consists of more than ten cup-like graphitic layers that can be clearly seen in FIG. 3B as indicated by the dashed lines. The resolution and contrast of FIGs. 3B and 3C are limited since the electron beam needs to penetrate through hundreds of nanometer thick CNF or Si-CNF hybrid^ but the structural characteristics are consistent with the high-resolution TEM studies using smaller CNFs in literature. This unique structure generated clusters of broken graphitic edges along the CNF sidewall which cause varied nucleation rates during Si deposition and thus modulate the density of the Si Layer 115 on the CNF 110 sidewall. The modulated density results in the ultra-high surface area Si structures indicated by a (100 nrn square) Box 310 in FIG 3A. The feather like Si structures of Si Layer 115 provide an excellent Li ion interface that results in very high Li capacity and also fast electron transfer to CNF 110. In FIG. 3 A the dark area at Tip 120 is Nickel catalyst for growth of the CNFs, Other catalysts can also be used.

[0035] FIGs, 3B and 3C are images recorded before (3B) and after (3C) lithium internaialion/extraction cycles. The sample in 3C was in the dclithiated (discharged) staie when ii was taken out of an electrochemical ceil. The dashed lines in FIG. 313 arc visual guidance of the stacked-cone graphic layers inside the CNFs 110. The long dashed lines in FIG. 3C represent the sidewall surface of the CNF 110.

[0036] As discussed elsewhere herein, the stacked-cone structure of CNFs 110 is drastically different from commonly used carbon nanotubes (CNTs) or graphite. The stacked-cone structure results in improved ΤΓ insertion, even w ithout the addition of Si Layer 115, relative to standard carbon nanotubes or nanowires. For example, the stacked-cone graphitic structure of CNFs 110 allows Li intercalation into the graphitic layers through the sidewall of CNFs 110 (rather than merely at the ends). The Li+ transport path across the wall of each of CNFs 110 is very short (with D -290 nm in some embodiments), quite different from the long path from the open ends in commonly used seamless carbon nanotubes (CNTs). FIG. 4 illustrates a schematic of the stacked-cone structure of CNFs 110. In this particular embodiment the average values of the parameters are: CNF radius tcnf= 74 nm, CNF wall thickness tw = -50 nm, graphitic cone angle Θ = 10°, and the graphitic cone length D = tw/sinO = 290 nm.

[0037] FIGs. 5A-5C illustrate an electrochemical characterization of -3 pm long CNFs 110. This characterization illustrates the phenomenon described in relation to FIG 4. FIG. 5A shows cyclic voltammograms (CV) from 1.5 V to 0.001 V versus a Li/Li+ reference electrode at 0.1, 0.5 and 1.0 mV/s scan rates. A lithium disk was used as the counter electrode. Data were taken from the second cycle and normalized to the exposed geometric surface area. FIG. 5B shows the galvanostatic charge-discharge profiles at C/0.5, Cl and C/2 power rates, corresponding to current densities of 647, 323 and 162 mA/g (normalized to estimated carbon mass) or 71.0, 35.5 and 17.8 pA/cm2 (normalized to the geometric surface area), respectively. FIG. 5C shows intercalation and extraction capacities (to left vertical axis) and Coulombic efficiency (to right vertical axis) versus the cycle number at C/1 charge-discharge rate. (The C/1 discharge rate = 1 hour, C/2 discharge rate = 120 min, 2C = C/0.5 = 30 min, etc.) [0038] A freshly assembled half-cell typically showed the open circuit potential (OCP) of the uncoated CNFs 110 anode was -2.50 to 3.00 V vs. Li/Li* reference electrode. The CVs measured between 0.001 V and 1.50 V show that Li* intercalation starts as the electropotential is below 1.20 V. The first cycle from OCP to 0.001 V involved the formation of a necessary protective layer, i.e. the solid electrolyte interphase (SEI), by the decomposition of solvent, salts, and impurities and thus presented a large cathodic current. Subsequent CVs showed smaller but more stable currents. The cathodic current associated with Li intercalation rose slowly as the electrode potential was swept to negative until a sharp cathodic peak appeared at 0.18 V. As the electrode potential was reversed to positive after reaching the low limit at 0.001 V, lithium extraction was observed in the whole range up to 1.50 V, indicated by the continuous anodic current and a broad peak at 1.06 V.

[0039] The CV features of CNF arrays 100 were somewhat different from those of staged intercalation into graphite and slow Li+ diffusion into the hollow channel of CNTs. Li-ion insertion into CNFs 110 is likely through intercalation between graphitic layers from the sidewall due to its unique structure. The TEM image in FIG. 3C indicates that the graphitic stacks in the stacked-cones inside the CNF 110 are somewhat disrupted during Li+ intercalation extraction cycles, likely due to the large volume change that occurs on Li+ intercalation. Some debris and nanoparticles are observed as white objects inside CNFs 110 as well as at the exterior surface.

[0040] The galvanostatic charge discharge profiles in FIG. 5B showed that the Li+ storage capacity decreased as the power rate was increased from C/2 to C/0.5 (C/0.5 is also referred to as “2C”). To make it easier to compare the rates (particularly for those higher than C/1), we use the fractional notation C/0.5 herein instead of “2C” that is more popularly used in the literature. The Li intercalation and extraction capacities were normalized to the estimated mass of the CNFs 110 (1.1 x 104 g em2) that was calculated based on a hollow vertically aligned CNF structure with the following average parameters: length (3.0 pm), density (1.1 x 10 CNFs per cm*), outer diameter (147 nm), and hollow inner diameter (49 nm, -1/3 of the outer diameter). The density of the solid graphitic wall of the CNFs 110 was assumed to be

•J the same as graphite (2.2 g/cm ). At the normal C/2 rate, the intercalation capacity was 430 mA h g’1 and the extraction capacity is 390 mA h g'1, both of which are slightly higher than the theoretical value of 372 mA h gu for graphite, which may be attributed to SEI formation and the irreversible Li+ insertion into the hollow compartments inside the CNFs 110. The extraction capacities were found to be more than 90% of the intercalation values at all power rates and both the intercalation and extraction capacities decreased by -9% as the power rate increased from C/2 to C/1 and by ~20% from C/1 to C/0.5, comparable to graphite anodes. [0041] Upon charge-discharge cycling, the intercalation capacity was found to slightly drop from 410 mA h g"1 to 370 mA h g’1 after 20 cycles at the C/1 rate, while the extraction capacity was maintained between 375 and 355 mA h g'1. The overall coulombic efficiency (i.e. the ratio of extraction capacity to intercalation capacity) was -94%, except in the first two cycles due to SEI formation on the CNF 110 surface. The SEI film is known to form readily on carbonaceous anodes during the initial cycles which allows lithium ion diffusion but is electrically insulating, leading to an increase in series resistance. The TEM image (FIG. 3C) and SEM image (FIG. 6A) show that a non-uniform thin film was deposited on the CNF 110 surface during charge-discharge cycles. In some embodiments, the SEI serves as a sheath to increase the mechanical strength of the CNFs 110, preventing them from collapsing into microbundles by the cohesive capillary force of a solvent as observed in the study with other polymer coatings.

[0042] FIGs. 6A-6C illustrates scanning electron microscopy images of 3 pm long CNFs 110, according to various embodiments of the invention. FIG. 6A shows CNFs 110 in delithiated (discharged) state after intercalation/extraction cycles. FIG. 6B shows CNFs 110 including Si Layer 115 after 100 cycles in the delithiated state. FIG. 6C shows CNFs 110 including Si Layer 115 after 100 cycles in the lithiated state. These images are 45 degree perspective views.

[0043] FIGs. 7A-7C illustrate results obtained using CNFs 110 including a Si Layer 115 as Li-ion battery anodes. These results were obtained using a nominal Si thickness of 0.50 pm. FIG. 7A shows cyclic voltammograms between 1.5 V and 0.05 V versus Li Li at 0.10, 0.50 and 1.0 mV s'1 scan rates. The measurements were made after the sample going through 150 charge discharge cycles and the data of the second cycle at each scan rate are shown. FIG. 7B shows galvanostatic charge discharge profiles at C/0.5, C/1 and C/2 power rates with the sample at 120 cycles. All profiles were taken from the second cycle at each rate. FIG. 7C shows insertion and extraction capacities (to the left vertical axis) and coulombic efficiency (to the right vertical axis) of two CNF Arrays 100 (used as electrodes) versus the chargedischarge cycle number. The first CNF Array 100 was first conditioned with one cycle at the C/10 rate, one cycle at the C/5 rate, and two cycles at the C/2 rate. It was then tested at the C/2 insertion rate and C/5 extraction rate for the rest of the 96 cycles. The filled and open squares represent the insertion and extraction capacities, respectively. The second electrode was first conditioned with two cycles each at C/10, C/5, C/2, C/1, C/0.5 and C/0.2 rates. It was subsequently tested at the C/1 rate for the next 88 cycles. The columbic efficiencies of both electrodes are represented by filled (1 st electrode) and open (2nd electrode) diamonds, which mostly overlap at 99%.

[0044] The CVs in FIG. 7A present very similar features to those of Si nano-wires. Compared to uncoated CNF Array 110, both the cathodic wave for Li+ insertion and the anodic wave for Li extraction shift to lower values (below ~0.5 and 0.7 V, respectively). The peak current density increases by 10 to 30 times after application of Si Layer 115 and is directly proportional to the scan rate. Clearly, alloy-forming Lf insertion into Si is much faster than intercalation into uncoated CNFs, which was limited by the slow diffusion of Li between graphitic layers. The cathodic peak at -0.28 V was not observed in previous studies on pure Si nanowires. The three anodic peaks representing the transformation of the Li-Si alloy into amorphous Si are similar to those with Si nanowires despite shifting to lower potentials by 100 to 200 mV.

[0045] The galvanostatic charge-discharge profiles of a CNF Array including Si Layer 115, shown in FIG. 7B included two remarkable features: (1) a high LL insertion (charge) and extraction (discharge) capacity of -3000 mA h (gsi)’1 was obtained at the C/2 rate even after 120 cycles; and (2) the Li+ capacity was nearly the same at the C/2, C/1, and C/0.5 power rates. In other words, the capacity of CNF Array 100 to operate as an electrode did not decline when charging rates were increased from C/2 to C/1 and C/0.5. Over these charging rates the capacity was nearly independent of charging rate, in various embodiments. The total Li+ storage capacity of CNF Arrays 100 including Si Layer 115 was about 10 times greater than CNF Arrays 100 that lacked Si Layer 115. This occurred even though the low potential limit for the charging cycle was increased from 0.001 V to 0.050 V. As a result, the amount of Li+ intercalation into the CNF core appears to have been negligible. The specific capacity was calculated by dividing only the mass of Si that was calculated from the measured nominal thickness and a bulk density of 2.33 g cm' . This method was chosen as an appropriate metric to compare the specific capacity of the Si Layer 115 to the theoretical value of bulk Si. For the 3.0 pm long CNFs 110 deposited with a Si Layer 115 of 0.456 pm nominal thickness, the real mass density of Si Layer 115 was -1.06 x 10'4 g cm'2, comparable to that of CNFs 110 (-1.1 x 10'4 g cm'2). The corresponding coulombic efficiency in FIG. 7B is greater than 99% at all three power rates, much higher than that of the CNFs 110 without Si Layer 115.

[0046] FIG. 8 illustrates how the capacity of CNF Array 100 varies with charging rate, according to various embodiments of the invention. Data is shown for several numbers of cycles. FIG. 8 shows average specific discharge capacity for a group of cycles with identical current rates versus the charge rate (C-rate) required to achieve full capacity in set hours (C/h e.g., full Capacity/hours). Vertical Lines are focused on C 4. 1C, 3C and 8C. The CNF Array 100 was first conditioned with two cycles each at C/8, C/4, C/2, C/1, C/0.8, C/0.4, and C/0.16 rates symmetrically, and subsequently tested at a C/1 symmetric rate for the next 88 cycles. This was repeated from cycle 101 to cycle 200. Starting at cycle 201, the electrode was cycled for five cycles at each of C/4, C/3, C/2, C/1, C/0.75, C/0.66, C/0.50, C/0.33, C/0.25, C/0.20 and C/0.15 rates symmetrically and subsequently tested at a C/1 symmetric rate for the next 45 cycles. This was repeated from cycle 301 to cycle 400 and from cycle 401 to cycle 500. The change in capacity is small (<16%) while the C-rate is varied by 32 fold. The electrode after 100 cycles showed increased capacity when the C-rate is changed from 3C to 8C. Thus, faster charge rates resulted in improved capacity. High capacity (>2,700 mAh/g) was obtained at both high and lower rates (C/4 and 8C). Capacity at rates above 3C increase as C-rate increased. The drop in specific capacity with the number of cycles is due to known, correctable, factors.

[00471 Both the CVs and charge discharge measurements indicated that the Li+ insertion into Si Layer 115 was fast and highly reversible, which are features desired for high-performance Li-ion battery anodes. This was further demonstrated (See FIG. 7C) with two long cycling tests on two identical samples at different testing conditions: (1) slow asymmetric tests with the C/2 rate for insertion and the C/5 rate for extraction; and (2) the fast symmetric test at the C/1 rate for both insertion and extraction. Both sets of data showed >98% coulombic efficiency over the long cycling except for the initial conditioning cycles (4 cycles in the former and 12 cycles in the latter at varied low rates). In the slow asymmetric tests, the insertion capacity only dropped by 8.3% from 3643 mA h g~* at the 5th cycle to 3341 mA h g' 1 at the 100th cycle. Even at the C/1 charge discharge rate, the insertion capacity only drops by 11% from 3096 mA h g’1 at the 13th cycle to 2752 mA h g’1 at the 100th cycle. The difference in the Li" capacity between these two sets of data was mostly attributable to the initial conditioning parameters and small sample-to-sample variations. This was indicated by the similar values of insertion-extraction capacity during the first few conditioning cycles in FIG. 7C at C/10 and C/5 rates. The faster rates (C/0.5 for 9th and 10th cycles and C/0.2 for 11th and 12th cycles in sample #2) were found to be harmful and caused an irreversible drop in the capacity. However, the electrode became stabilized after longer cycling. As shown in FIG. 7B, the charge-discharge profiles are almost identical at C/2, C/1, and C/0.5 rates, which were measured with sample #1 after going through 120 cycles. This is over a charging rate variation of four times.

[0048] The specific capacity of the Si Layer 115 in the range of 3000 to 3650 mA h g'1 is consistent with the highest values of amorphous Si anodes summarized in literature. It is remarkable that the entire Si shell in the CNF Array 110 was active for Li+ insertion and remained nearly 90% of the capacity over 120 cycles, which to our knowledge has not been achieved before except with flat ultrathin (<50 nm) Si films. The specific capacity disclosed herein is significantly higher than those reported using other nanostructured Si materials at similar power rates, including -2500 mA h g'1 at the C/2 rate and -2200 mA h g'1 at the C/1 rate with Si NWs, and -800 mA h g'1 at the C/1 rate with randomly oriented carbon nanofiber-Si core-shell NWs. Clearly, the coaxial core-shell NW structure on well-separated CNFs 110, such as included in various embodiments of the invention, provides an enhanced charge-discharge rate, nearly full Li storage capacity of Si, and a long cycle life, relative to the prior art.

[0049] As shown in FIG. 7C, an anomalously high insertion capacity (-4500 mA h g'1) was always observed in the initial cycles, which was 20-30% higher than the latter cycles. In contrast, the extraction values were relatively stable over all cycles. The extra insertion capacity can be attributed to the combination of three irreversible reactions: (1) the formation of a thin SEI (surface electrolyte interphase) layer (of tens of nanometers); (2) reactions of Li with SiOx presented on the Si surface (SiOx + 2xLi -> Si + xLioO); and (3) the conversion of the starting crystalline Si coating with a higher theoretical capacity (-4200 mA h g1) into amorphous Si with lower capacity (<3800 mA h g’1). The TEM image (FIG. 3C) and SEM image (FIG. 6B) showed that a non-uniform SEI can be deposited on the surface of Si Layer 115 after the charge-discharge cycles. This elastic SEI film can help secure Si Layer 115 on the CNF 110 surfaces as CNF Array 110 goes through the large volume expansioncontraction cycles that occur during the charge-discharge cycles. The dramatic difference between the SEM images in FIGs. 6B and 6C indicates the large expansion of Si Layer 115 in the lithiated (charged) state relative to the non-lithiated state. (Although some of the expansion may be due to oxidation of Li by air as the electrochemical cell was dissembled for imaging.) Note that the production of SEI during initial charge-discharge cycles causes the differences seen in Si Layer 115 between FIGs. 3A and 3B. In FIG. 3B the Si has interacted with electrolyte to produce SEI that fills the gaps between the feather-like structures. The interaction can include mixing, chemical reactions, charge coupling, encapsulation, and/or the like. The Si Layer 115, therefore, looks more uniform in FIG. 3B. However, the Si Layer 115 now comprises interleaved layers of Si (the feather-like structures) and SEI. Each of these interleaved layers can be on the order of a few 10s of nanometers. The SEI layer can be an ion permeable material that is a product of interaction between the electrolyte and Si Layer 115 (or other electrode material).

[0050] The crystalline and amorphous structure of the Si shell was revealed by Raman spectroscopy. As shown in FIG. 9, the pristine CNF Array 100 including Si Layer 115 showed multiple broad bands overlapped in the range of 350 to 550 cm'1 corresponding to amorphous Si, and a much higher sharp band at 480 cm'1 corresponding to nanocrystalline Si. After charge-discharge tests, the sharp peak disappeared while the broad bands merged into a single peak at 470 cm’1. The bare CNFs 110 did not show any feature in this range. The crystalline Si peak downshifted by -40 cm'1 from that measured with a single-crystalline Si(100) wafer and by -20 to 30 cm’1 from other micro-crystalline Si materials. This shift was likely due to the much smaller crystal size and large disorders. The original Si Layer 115 likely consisted of nanocrystals embedded in an amorphous matrix associated with the feather-like TEM image in FIG. 3A. After initial cycles, the Si nanocrystals were converted into amorphous Si, consistent with the TEM images after the cycling test (see FIGs. 3B and 3C). However, the Si Layer 115 apparently did not slide along the CNF, in contrast to the large longitudinal expansion (by up to 100%) in pure Si NWs. Si Layer 115 was, thus, securely attached to CNFs 110 for over 120 cycles. The volume change of the Si shell during Li+ insertion was dominated by radial expansion, while the CNF-Si interface remained intact. [0051] Various embodiments of the invention include CNFs 110 having different lengths and silicon shell thickness. One factor that can be controlled when CNFs 110 are generated is the open space between each CNF 110, e.g., the mean distance between CNFs 110 within CNF Array 100. This space allows Si Layer 115 to expand radially when charging and, thus in some embodiments provides stability. Because an optimum electrode structure depends on both the length of CNFs 110 and the thickness of Si Layer 115, it is sometimes desirable to use longer CNFs 110 and thicker Si Layers 115 in order to obtain higher total Li ’ storage capacity. Longer CNFs 110 do correlate with greater storage capacity. FIGs 10A-10C shows the variation of Li insertion-extraction capacities and the coulombic efficiency over 15 charge-discharge cycles with three 10 pm long CNF 110 samples deposited with Si Layer 115 at a nominal thickness of 0.50, 1.5 and 4.0 pm, respectively. After conditioning at the C/10 rate for the first cycle and the C/5 rate for the second cycle, asymmetric rates (C/2 for insertion and C/5 for extraction) were used in subsequent cycles similar to the measurements of sample #1 in FIG. 7C. This protocol provided nearly 100% coulombic efficiency and minimum degradation over the cycles. The nominal thickness was measured in situ with a quartz crystal microbalance during sputtering.

[0052] The specific capacities as high as 3597 mA h g'! and 3416 mA h g’1 were obtained with 0.50 and 1.5 pm thick Si Layer 115, respectively, very similar to that with 0.50 pm thick Si Layer 115 on 3.0 pm long CNFs 110 (see FIG. 7C). The capacity remained nearly constant over 15 cycles. However, the electrode with 4.0 pm nominal Si thickness showed a significantly lower specific capacity at only 2221 mA h g'1. This indicates that, with expansion, the Si Layers 115 from adjacent CNFs 110 began to contact into each other, limiting them from further expansion and limiting diffusion of Li between CNFs 110. As a result, only a fraction of the silicon coating was active in lithium insertion. The cycle stability was correspondingly worse than the samples with thinner Si Layers 115.

[0053] The same amount of Si (500 nm nominal thickness) on CNF Arrays 110 comprising 10 pm long CNFs 110 gave nearly the same amount of Li+ storage capacity (3597 mA h g’1, see Fig. 6a) as that of 3 pm long CNFs 110 (3643 mA h g'1, see FIG. 7C), even though the carbon mass is more than 3 times higher. This is very strong evidence that the contribution of CNFs 110 is negligible in calculating Li+ storage. It is likely that very little Li+ ions were intercalated into CNFs 110 in the Si-coated sample, this contributes to the stability of the structure during multiple charge-discharge cycles.

[0054] The variation of the specific Li+ storage capacity in the three samples correlated well with their structures revealed by the SEM images illustrated in FIGs. 11 A-l 1C. FIGs. 11A-11C show scanning electron microscopy images of freshly prepared CNF Arrays 100 (on -10 pm long CNFs 110). The Si Layer 115 was generated using a nominal Si thickness of (a) 0.50 pm, (b) 1.5 pm, and c) 4.0 pm, which were measured in-situ using a quartz crystal microbalance during deposition. All images are 45° perspective views. At 0.50 pm nominal Si thickness, the average tip diameter was found to be -388 nm on the 10 pm long CNFs, much smaller than the -457 nm average diameter on the 3.0 pm long CNFs 110. The Si

Layer 115 was thinner but more uniformly spread along the 10 pm long CNFs 110.

[0055] It is noted that growing 10 pm CNFs 110 took 120 min, about six times as long as growing the 3 pm CNFs 110. Some Ni Catalysts were slowly etched by NHj during the long PECVD process, resulting in continuous reduction in the Ni nanoparticle size and leading to the tapered Tip 120 (as shown in FIG. 12). The CNF 110 length variation also increased with long CNFs 110. These factors collectively reduced the shadow effects of the Tip 120. As a result, even at 1.5 pm nominal Si thickness, the CNFs 110 coated with Si Layer 115 are well separated from each other. The SEM image of 1.5 pm Si on 10 pm CNF Arrays 100 (FIG. 1 IB) is very similar to that of 0.50 pm Si on 3.0 pm CNF Arrays 110 (FIG. 2B). But as the nominal Si thickness was increased to 4.0 pm, the Si Layers 115 clearly merged with each other and filled up most of the space between the CNFs 110 (see FIG. 10C). This reduced the free space needed to accommodate the volumetric expansion of the Si Layer 1151. As a result, the specific Li+ storage capacity significantly dropped.

[0056] FIGs. 11A and 1 IB each include roughly the same number of CNFs 110, however, in FIG. 1 IB has substantially fewer visible Tips 120. This is because Si Layer 115 can form a nano fiber/silicon complex that includes a single CNF 110 (a cross-section of which is shown in FIG. IA). Or, Si Layer 115 can form a nanofiber/silicon complex that includes two, three or more CNF 110 under a single cover of silicon. This occurs when two or more CNFs 110 come together during the Si Layer 115 deposition process. A nanofiber/silicon complex is a structure that includes a continuous Si Layer 115 that envelops one or more CNF 110. A cross-section of a nanofiber/silicon complex that includes two CNF 110 is illustrated in FIG. 1 ID. In various embodiments at least 1%, 5% or 10% of nanofiber/silicon complexes include more than one CNF 110.

[0057] In various embodiments, instances of CNF Arrays 100 having 0.50 and 1.5 pm nominal Si thicknesses have comparable mass-specific capacities of 3208 ± 343 and 3212 ± 234 mA h g'1, respectively. The samples with a 4.0 pm nominal Si thickness give much lower capacity at 2072 ± 298 mA h g'1. The thinner Si coatings are fully activated and provide the maximum Li insertion capacity that amorphous Si could afford. On the other hand, the areaspecific capacity increases proportionally with the Si thickness from 0.373 ± 0.040 mA h cm' 7 . 9 " at 0.50 pm Si to 1.12 ± 0.08 mA h cm'" at 1.5 pm Si thickness, but drops off from the linear curve to give 1.93 ± 0.28 mA h cm'2 at 4.0 pm nominal Si thickness. Clearly, at this thickness, only a fraction of the extra silicon in the thick Si coating is actively involved in Li storage. The thickness of 4.0 pm is greater than the mean distance between CNFs 110. The electrochemical results are consistent with the structure shown in SEM image in FIG. 11C, which shows that space between CNFs 110 is essentially filled.

[0058] In various embodiments of the invention, the structure of CNF Array 100 includes an Si Layer of approximately 200 to 300 nm radial thickness on CNFs 110 having a length of approximately 30-40, 40-75, 75-125 microns (or more or combinations thereof) and diameters on the order of -50 nm. In some embodiments, these CNF Array 100 are grown on conductive foils having a thickness within the ranges of -10 microns, -10-20 microns, -10-50 microns, or more. In various embodiments, Si (equivalent to 1.5 pm nominal thickness on a flat surface) is deposited onto 10 pm long CNFs 100 to form CNF Arrays 100. This is accomplished while maintain the open vertical core-shell nanowire structure with individual CNFs 110 well separated from each other such that Li ions can penetrate the CNF Arrays 100 between the CNFs 110. This unique hybrid architecture allowed the Si Layers 115 to freely expand/contract in the radial direction during Li+ insertion and extraction. High-performance Li storage with a mass-specific capacity of 3000 to 3650 mA h g'1 was obtained even at the C/1 rate. The capacity matched the maximum value that would be expected from a similar mass of amorphous Si, indicating that the Si Layer 115 was fully active. This 3D nanostmctured architecture enables effective electrical connection with bulk quantities of Si material while maintaining a short LL- insertion extraction path. Asa result, high capacity near the theoretical limit is possible for over 120 charge discharge cycle. There was little change in capacity as ihe rate was increased 20 times from C/10 to C O.5 (or 2C). The high capacity at significantly improved charging and power rates and the extraordinary cycle stability make this novel structure a choice anode material for high-performance Li-ion batteries. The same core-shell concept may be applied to cathode materials by replacing the Si shell with TiOj, LiCoO-j. LiNiCL, LiMibOi, LiFei*04, or the like. (0059] FIG. 13 illustrates methods of producing the CNF Arrays 100 disclosed herein. In a Provide Substrate Step 1310a Substrate 105 suitable for growth of CNFs 110 is provided. Substrate 105 may include a variety of materials, for example Cu. Substrate 105 is optionally a conductive foil having a thickness described elsewhere herein. In an optional Provide Nucleation Sites Step 1320 nucleation cites for the growth of CNFs 110 are provided on Substrate 105. A variety of nucleation materials, such as Ni particles, are known in the art. The nucleation cites are optionally provided at a density so as to produce mean distances between CNFs 110, such as those taught elsewhere herein. Provide Nucleation Sites Step 1320 is optional in embodiments in which nucleation is not required for growth of CNFs 110, or similar structures.

[0060] In a Grow CNFs Step 1330 CNFs 110 are grown on Substrate 105. The CNFs 110 are grown io produce the stacked-cone structure taught elsewhere herein.

The CNFs 1 10 can be grown to any of the lengths taught elsewhere herein. Growth is optionally accomplished using PECVD processes such as those taught or cited in “.A high-performance lithium-ion battery anode based on the core-shell heterostruciure of silicon-coated vertically aligned carbon nanofibers'* Klankowski el al J. Mater. Client. .4, 2013, 1. 1055. 10061 ] hi an Apply Si Layer Step 1340 an intercalation material such as Si Layer 115 is applied to the grown CNFs 110. The applied material may have any of the nominal thicknesses taught elsewhere herein so as to produce a Si Layer 115 thickness of tens or hundreds of nanometers. In an optional Condition Step 1350 the CNF Array 100 produced using Steps 1310-1340 is conditioned using one or more lithium intercalation cycles, [9062] Several embodiments are specifically illustrated and/or described herein. However, it will be appreciated that modifications and variations are covered by the above teachings and within the scope of the appended claims without departing from the spirit and intended scope thereof. For example, while a Cu substrate and Li charge carriers are discussed herein other substrates and charge carriers will be apparent to one of ordinary skill in the art. Si Layer 115 is optionally formed of intercalation materials in addition to or as an alternative to silicon. For example, tin, germanium, carbon, silicon, or combinations thereof could be used as intercalation material.

[0063] The electrodes taught herein may be included in a wide variety of energy storage devices including capacitors, batteries and hybrids thereof. These energy storage devices will be used in, for example, load balancing devices, communication devices, backup power supplies, vehicles and computing devices.

Claims

1. An energy storage system comprising: a conductive substrate; a plurality of aligned carbon nanofibers grown on the substrate with a layer of intercalation material disposed on the carbon nanofibers, the carbon nanofibers each including a plurality of multi-walled carbon nanotubes, wherein the carbon nanotubes are disposed such that ion intercalation can occur between walls of the nanotubes through the sidewalls of the carbon nanofibers, with the carbon nanofibers including a stacked-cone structure; and an electrolyte including one or more charge carriers.

2. The system of claim 1, the layer of intercalation material having a feather-like structure resulting from the stacked-cone structure.

3. The system of claim 1 or 2, the layer of intercalation material comprising feather-like structures of silicon filled with surface electrolyte interphase.

4. The system of any of claims 1-3, the intercalation material having a nominal thickness of between 0.01 and 25 pm.

5. The system of any of claims 1-4, wherein the layer of intercalation material includes nano fiber/intercalation material complexes, some of the nanofiber/ intercalation material complexes including one nanofiber and some of the nanofiber/ intercalation material complexes including two nanofibers.

6. The system of any of claims 1-5, the layer of intercalation material on the carbon nanofibers comprising silicon, the silicon having a nominal thickness of between 0.01 pm and 25 pm.

7. The system of any of claims 1-6, wherein the carbon nanofibers are between 3.0 and 200 pm in length.

8. The system of claim 1, wherein the intercalation material includes silicon.