Fabricating Genetically Engineered High-Power Lithium-Ion Batteries Using Multiple Virus Genes

See allHide authors and affiliations

Science  22 May 2009:
Vol. 324, Issue 5930, pp. 1051-1055
DOI: 10.1126/science.1171541


Development of materials that deliver more energy at high rates is important for high-power applications, including portable electronic devices and hybrid electric vehicles. For lithium-ion (Li+) batteries, reducing material dimensions can boost Li+ ion and electron transfer in nanostructured electrodes. By manipulating two genes, we equipped viruses with peptide groups having affinity for single-walled carbon nanotubes (SWNTs) on one end and peptides capable of nucleating amorphous iron phosphate(a-FePO4) fused to the viral major coat protein. The virus clone with the greatest affinity toward SWNTs enabled power performance of a-FePO4 comparable to that of crystalline lithium iron phosphate (c-LiFePO4) and showed excellent capacity retention upon cycling at 1C. This environmentally benign low-temperature biological scaffold could facilitate fabrication of electrodes from materials previously excluded because of extremely low electronic conductivity.

Lithium-ion battery electrodes store and release electrical energy by insertion and extraction of Li+ ions and electrons through the electrode materials. Therefore, increasing transport of Li+ ions and electrons in electrodes can enhance energy storage at high charge and discharge rates. Controlling nanostructure has become a critical process in developing electrode materials to boost transport in composite electrodes (1, 2), especially for the electrically insulating transition metal phosphate cathode materials. Among them, iron phosphate–based materials have elicited attention as promising Li+-ion battery positive electrode materials due to their lower toxicity, lower cost, and improved safety through improved chemical, thermal, and structural stability for high-power applications (3). However, their practical use has been constrained due to kinetic limitations, which result in poor charge- and discharge-rate capability and fading of capacity upon prolonged cycling. To address the rate limitation of these materials, most researchers have focused on tailoring particle size (4, 5) to reduce both the ionic and electronic path within the particles and enhancing electronic conductivity with surface carbon-coating layers (6) or conducting nanoparticles additives (7, 8). However, the fabrication of nanosized particles is still challenging because the materials require at least 350°C for crystallization and carbon coating. Despite the recent advances in synthesis methods, the smallest particle size remains 20 to 40 nm (5).

Biological systems offer capabilities for environmentally benign materials synthesis. An M13 virus–based biological toolkit has been developed for the design of nanoarchitectured structures and materials (912). Our group has shown that M13 bacteriophage (phage or virus) can be used for battery device fabrication with improved performance by synthesizing electrochemically active anode nanowires and organizing the virus on a polymer surface (11, 13). However, in designing nanostructured electrodes with better electrical wiring for high-power batteries, multifunctionality of the virus is required. Multifunctional viruses have been engineered with desired modifications on different positions of the protein coat (10, 12, 14). Here we demonstrate a genetically programmed multifunctional virus as a versatile scaffold for the synthesis and assembly of materials for high-power batteries.

Virus-enabled nanostructured cathode materials were first demonstrated by templating amorphous anhydrous iron phosphate (a-FePO4) on the E4 virus. E4 is a modified M13 virus that has tetraglutamate (EEEE) fused to the N terminus of each copy of pVIII major coat protein. Due to the presence of extra carboxylic acid groups compared with wild-type M13 virus (M13KE), the E4 virus exhibits increased ionic interactions with cations and can serve as a template for materials growth (11, 13, 15). Because only one gene (gVIII in Fig. 2A) has been modified for the desired peptide motif on pVIII, we call this E4 clone a one-gene system. a-FePO4 nanowires were produced on silver nanoparticles (Ag NPs)–loaded E4 virus. Details of the synthesis procedure are given in the supporting online material (16). Loading of uniformly distributed Ag NPs along the coat protein of E4 virus (fig. S1) was initially intended to increase electronic conductivity (7, 8, 11). The chemical analysis by direct current plasma atomic emission spectroscopy confirmed the atomic ratio of Fe to P as 1:1. Although viruses themselves have phosphate groups in their DNA (7270 phosphate groups per virus particle), the fraction of phosphate groups from DNA is <1%. Figure 1A shows transmission electron microscope (TEM) images of a-FePO4 nanowires with particle sizes of 10 to 20 nm in diameter templated on the virus. Generally, hydrated a-FePO4 (a-FePO4·nH2O, n = 2 to 4) is precipitated in aqueous solutions containing Fe3+ and PO43− ions around pH = 7 to 8, and anhydrous structures can be obtained through the dehydration of a-FePO4·nH2O by thermal annealing at 400°C. Most structural water in a-FePO4·nH2O is removed from the structure around 200°C (17). Surprisingly, the viral nanowires produced on Ag NP-loaded E4 were anhydrous as synthesized, as shown by thermogravimetric analysis (TGA) (Fig. 1B and fig. S2). Without Ag NPs, nanowires have about 10 weight percent (wt %) structural water, which corresponds to n = 1 in a-FePO4·nH2O. X-ray powder diffraction of a-FePO4 nanowires on Ag NPs-loaded E4 (fig. S3A) showed only peaks indexed as silver chloride (AgCl). We speculate that the dehydration of FePO4·nH2O is related to the chlorination of Ag NPs, which could occur during the incubation with the iron chloride precursor. Part of the chlorinated AgCl was reduced to metallic Ag after electrochemical test (fig. S3B). The reduced metallic Ag could enhance local electronic conductivity as Au nanoparticles could in Co3O4/Au heterostructured nanowires (11). Although the exact mechanism of dehydration is under investigation, dehydration of structural water without thermal treatment was accomplished by low-temperature and environmentally benign chemistry. The dehydrated structure increases the theoretical capacity to 178 mAh/g, making it a good cathode material.

Fig. 1

Characterization of a-FePO4 nanowire cathodes in a one-gene viral system (E4). (A) TEM images of templated a-FePO4 nanowires on E4 viruses. (Inset) Magnified images of the same nanowires. (B) TGA curve of a-FePO4 nanowires synthesized on Ag NP–loaded E4. For comparison, a TGA curve of a-FePO4·H2O grown on E4 virus (without Ag NPs) is also presented. (C and D) Electrochemical performance of a-FePO4 viral nanowires on E4 tested between 2.0 and 4.3 V. Active materials loading was 2.63 mg/cm2. (C) First discharge curves at different rates. (D) The Ragone plot representing rate performance in terms of specific power versus specific energy (only the active electrode mass is included in the weight).

Fig. 2

Biological toolkits: genetic engineering and biomolecular recognition. (A) A schematic presentation of the multifunctional M13 virus is shown with the proteins genetically engineered in this study. The gene VIII protein (pVIII), a major capsid protein of the virus, is modified to serve as a template for a-FePO4 growth, and the gene III protein (pIII) is further engineered to have a binding affinity for SWNTs. (B) A schematic diagram for fabricating genetically engineered high-power lithium-ion battery cathodes using multifunctional viruses (two-gene system) and a photograph of the battery used to power a green LED. The biomolecular recognition and attachment to conducting SWNT networks make efficient electrical nanoscale wiring to the active nanomaterials, enabling high power performance. These hybrid materials were assembled as a positive electrode in a lithium-ion battery using lithium metal foil as a negative electrode to power a green LED. Active cathode materials loading was 3.21 mg/cm2. The 2016 Coin Cell, which is 2 cm in diameter and 1.6 mm in thickness, was used. LED power dissipation was 105 mW.

The electrochemical performance of viral a-FePO4 nanowires as a lithium-ion battery cathode was evaluated (Fig. 1, C and D). Positive electrodes were prepared by mixing viral a-FePO4 with Super P (TIMCAL, SUPER P Li) carbon black and polytetrafluoroethylene (PTFE) binder in a mass ratio of 70:25:5. Details of the weight ratio of components are given in the supporting online material (16). The first discharge capacity at a low discharge rate of C/10 (18) was 165 mAh/g (93% of the theoretical value) and that of 1C discharge rate was 110 mAh/g (Fig. 1C) (19). The rate performance is also presented as a Ragone plot (Fig. 1D). In most electrode materials, specific energy decreases substantially as one applies more power (high rates), drawing more current from the electrodes (20). These rate performance values are similar to the best reported values for a-FePO4 synthesized at high temperature (21). Even with this one-gene system, the nanostructuring of a-FePO4 nanowires by the virus enabled an enhanced performance. However, high-power performance and capacity retention upon cycling of both the biologically and traditionally synthesized electrodes are still inferior to commercially available c-LiFePO4 cathodes.

Because our particles were already 10 to 20 nm in diameter, our strategy for improved performance was to increase the electronic conductivity in the cathode by achieving better electrical contact between the active materials. Although metallic Ag NPs can locally enhance the electronic conductivity, more important for improved high-power performance is a percolating network throughout the electrodes. It is known that incorporation of well-dispersed materials with high conductivity and high aspect ratio leads to efficient percolating networks (22, 23). Carbon nanotubes (CNTs) have been shown to meet these needs (23); thus, well-dispersed single-walled CNTs (SWNTs) in water were used. However, conventional composite electrode fabrication processes inevitably suffer from aggregation of carbon particles, thereby diminishing contact with the active materials (23). To achieve better electrical wiring to our biologically derived a-FePO4, we engineered a specific affinity between the conducting material and active material.

Because the major coat protein of the E4 virus serves only as a template for a-FePO4 nanowire growth, additional genetic modification was required to engineer the E4 virus to have a binding affinity for SWNTs. In this context, the gene III protein (pIII), a minor coat protein located at one end of the virus (Fig. 2A), is an ideal tool because gene III can be controlled independently of gene VIII to insert foreign DNA encoding pIII-displayed peptides. Moreover, the peptide sequences identified through the phage display with a pIII phage-display library can be directly inserted into the E4 virus without losing functionality (12). Therefore, phage-display experiments to search for peptide sequences with a strong binding affinity for SWNTs were done first, followed by genetic engineering into the E4 virus to produce a multifunctional virus structure [see Methods for details of the procedure (16)]. Several consensus sequences were obtained from separate phage-display screening experiments. Among them, sequences N′-HGHPYQHLLRVL-C′ (24) (named MC#1) and N′-DMPRTTMSPPPR-C′ (MC#2) were selected for further experiments. The sequence MC#1 started with histidine (H), whose appearance in the first position was often observed in CNT-binding sequences (25). Also, it contained several aromatic residues (H and Y), which were expected to bind favorably to the graphene surface via π-stacking interaction (26). The sequence MC#2 is quite different from MC#1, and the binding affinity of clone MC#2 was approximately four times as high as that of clone MC#1, whose binding affinity was already two-and-a-half times as high as that of wild-type M13KE in the binding-affinity tests (27) (fig. S4, A and B). The strong binding of sequence MC#2 can be explained by the location of the hydrophobic segments of the sequence. The calculated hydrophobicity plot (fig. S5) shows a tri-block structure with hydrophilic regions on both ends and the hydrophobic region in the middle of the sequence. A tri-block structure of hydrophilic-hydrophobic-hydrophilic polymers was effective when used to suspend SWNTs (25, 28).

To genetically engineer E4 virus as a multifunctional biological platform, we fused the selected sequences, MC#1 and MC#2, independently onto the N-terminus of pIII of E4 virus, producing clones EC#1 and EC#2, respectively (16). Because two genes (gIII and gVIII in Fig. 2A) were engineered with the desired modification on both pIII and pVIII proteins, we called it a two-gene system.

A schematic diagram for constructing the genetically engineered high-power lithium-ion battery using the multifunctional two-gene virus system is illustrated in Fig. 2B. All viruses were loaded with Ag NPs to synthesize anhydrous a-FePO4. Formation of anhydrous a-FePO4 on pVIII preceded the interaction with SWNTs. The synthesis procedure of anhydrous a-FePO4 nanowires on the multifunctional viruses was the same for growth on the one-gene system. Viral a-FePO4 solutions were then incubated with the SWNT suspensions to form a-FePO4/SWNTs hybrid nanostructures (16). The photograph in Fig. 2B is the actual assembled lithium-ion battery powering a light-emitting diode (LED) using Li metal as a negative electrode. The virus-enabled high-power battery could power a green LED with a small amount of active materials loading of 3.21 mg/cm2. Although this cell was assembled with lithium foil as negative electrode, we have successfully made full virus-based 3-V batteries with various negative electrode materials (fig. S6).

The morphology of hybrid a-FePO4/SWNT nanowires on the EC#2 virus is shown in Fig. 3, C to E. In the high-resolution TEM image (Fig. 3E), six to eight SWNTs are bundled with diameters of 4 to 5 nm. The TEM images show that a-FePO4 nanowires templated on the multifunctional virus were tethered to SWNTs mainly through the pIII attachment; however, they made multiple contacts with neighboring SWNTs because of close positioning. To explore the effect of specificity, we also mixed the one-gene system (E4) viral nanowires solution with SWNTs. Most viral a-FePO4 nanowires on E4 did not make contact with SWNTs, and even if they did, the contact did not seem to be due to specific binding with SWNTs (fig. S7). Moreover, SWNTs aggregated by themselves when there was no specific binding on pIII, suggesting that SWNT-specific viruses enhanced dispersion of SWNTs in solution. Similar observation has been reported showing that SWNT-specific peptides can disperse SWNTs, whereas nonspecific peptides cannot (25).

Fig. 3

Morphology of the a-FePO4 grown on the multifunctional viruses/SWNT hybrid nanostructures. TEM images. (A) a-FePO4 nanowires templated on EC#2 viruses (before interacting with SWNTs). EC#2 virus is a two-gene system virus with the strongest binding affinity to SWNTs. (B) SWNTs only (before interacting with viral a-FePO4). (C to E) a-FePO4 grown on EC#2 attached to SWNTs. (C) Low magnification (×10,000). (D) Higher magnification (×30,000). (E) High-resolution TEM (HRTEM) images (×800,000). For HRTEM imaging, surfactants were removed by washing with acetone. Material-specific tethering of the viral a-FePO4 to the SWNTs is visualized. The amorphous nature of FePO4 was also confirmed.

The electrochemical properties of viral a-FePO4/SWNT hybrid materials with 5 wt % SWNTs were evaluated and compared (Fig. 4). Positive electrodes were prepared by mixing viral a-FePO4/SWNT hybrid composites with Super P carbon black and PTFE binder in a mass ratio of 90:5:5 (16). The addition of 5 wt % extra carbon was used to increase the total volume, making the powder easier to handle. Without extra carbon, the electrodes showed slightly higher polarization at high rates, but the difference was not substantial. As demonstrated in the first discharge profiles (see Fig. 4A and fig. S8A for full discharge/charge curves) (19), electrochemical performances improve markedly as the binding affinity to the SWNTs increases. Specific capacity at a low discharge rate of C/10 increased from 143 mAh/g (E4) to 160 mAh/g with EC#1 and to 170 mA·hour/g with EC#2. The performance improvement is more pronounced at higher rates. Discharge profiles of the two-gene system show much lower polarization and maintain much higher capacity than those of the one-gene system at high rates. When compared with the best reported capacity for a-FePO4 at a high rate of 3C (80 mAh/g) (21), EC#2 showed a capacity of 134 mAh/g, confirming substantially improved high-power performance. Moreover, when we cycled EC#2 between 1.5 and 4.3 V, the first discharge capacity at 10C reached 130 mAh/g. No published data for a-FePO4 are available for comparison at rates higher than 3C, but this capacity value obtained for the two-gene system is comparable to the capacity from state-of-the-art c-LiFePO4. The power performance of the multifunctional virus-based cathode was further compared with a Ragone plot. Figure 4B shows that two-gene system–based materials delivered much higher energy than the one-gene system at high power. At a specific power of 4000 W/kg (corresponding to a rate of ~10C), the energy density of EC#1 and EC#2 was two times and three times as high, respectively, as that of E4. Again, the high-power performance scales with binding affinity. In Fig. 4B (inset), the rate performance of E4 virus–based cathodes with either Super P carbon or SWNTs was tested. Well-dispersed SWNTs by themselves make better electrical wiring to active materials due to better percolation networks than carbon black powders (23), confirming the importance of nanoscale electrical wiring. Figure 4C shows the stable capacity retention of a-FePO4/SWNT hybrid electrodes upon cycling at 1C. Up to 50 cycles, virtually no capacity fade was observed. A slight capacity loss after the first cycle is a characteristic of a-FePO4 materials (17, 21). When cycled at C/10 rate again after the sample was tested for several cycles at rates from C/10 to 10C, the original capacity was recovered, confirming structural stability (fig. S8B). Structural stability of viral a-FePO4/SWNT hybrid nanostructures was induced by materials-specific binding and stiff, robust carbon nanotubes, leading to excellent retention at a low SWNT content of 5 wt %. Because the density of SWNTs is 1.33 g/cm3 (23), it would decrease the volumetric energy density of the hybrid electrodes. However, although we adopted SWNTs to show that we can achieve nanoscale wiring by genetic engineering, we expect that we could optimize the fraction of the conducting additives by using even better-conducting nanowires with high aspect ratio and higher density.

Fig. 4

Electrochemical properties of the a-FePO4 viral nanowires in two-gene systems tested between 2.0 and 4.3 V. Rate capabilities and capacity retention upon cycling of the a-FePO4 nanowires/SWNTs hybrid electrodes templated on different clones are presented. All a-FePO4/SWNT hybrid materials had 5 wt % SWNTs. EC#2 virus is a two-gene system virus with the strongest binding affinity to SWNTs; EC#1 is a two-gene system virus with moderate binding affinity; and E4 is a one-gene system virus with no insert on pIII. (A) First discharge curves at different rates. Active materials loading: E4, 2.34 mg/cm2; EC#1, 2.31 mg/cm2; and EC#2, 2.62 mg/cm2. (B) Ragone plot showing improvement in high-power performance with higher binding affinity toward SWNTs (only active electrode mass is included in the weight). (Inset) Comparison of rate capability of E4 virus–based cathodes with either Super P carbon or SWNTs. Electrodes with well-dispersed SWNTs, even with much smaller amounts, exhibited improved rate performance due to better percolation networks than carbon black powders. (C) Capacity retention for 50 cycles at 1C rate. There was no obvious fading for at least 50 cycles. Active materials loading: E4, 2.90 mg/cm2; EC#1, 2.22 mg/cm2; and EC#2, 2.27 mg/cm2.

There have been efforts to electrically address electrode materials with poor electronic conductivity through nanoscale wiring of active materials (8, 29, 30). However, the wiring tools used so far were functionalized for a single component, either active materials (8, 30) or conducting materials (29). The wiring did not completely exploit specificity but depended on the random occurrence of contacts between conducting networks and active materials. By developing a two-gene system with a universal handle to pick up electrically conducting carbon nanotubes, we devised a method to realize nanoscale electrical wiring for high-power lithium-ion batteries using basic biological principles. This biological scaffold could further extend possible sets of electrode materials by activating classes of materials that have been excluded because of their extremely low electronic conductivity.

Supporting Online Material

Materials and Methods

Figs. S1 to S8


  • * These authors contributed equally to this work.

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Rates are reported in C-rate convention, where C/n is the rate (current per gram) corresponding to complete charging or discharging to the theoretical capacity of the materials in n hours. Here, 1C corresponds to 178 mA/g.
  3. In the rate test, the cell was charged at C/10 rate to 4.3 V and then held at 4.3 V until the current density was lower than C/100 and discharged at different rates.
  4. Abbreviations for the amino acid residues are as follows: D, Asp; G, Gly; H, His; L, Leu; M, Met; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; and Y, Tyr.
  5. This work was supported by the Army Research Office Institute of the Institute of Collaborative Biotechnologies (ICB) and U.S. NSF through the Materials Research Science and Engineering Centers program. H.Y. is grateful for Korean Government Overseas Scholarship. W.-J.K. is grateful for support from the Korea Research Foundation Grant funded by the Korean Government (MOEHRD) (KRF-2005-214-D00260). K.K is grateful for funding support from Korea Science and Engineering Foundation of the Ministry of Education, Science and Technology (No. R01-2008-000-10913-0) and Energy Resource Technology Development program by the Ministry of Knowledge Economy (No. 2008-E-EL11-P-08-3-010). M.S.S. is grateful for funding from the NSF and the Office of Naval Research Young Investigator Grant.
View Abstract

Navigate This Article