Nine years ago DOE dedicated $2B primarily for manufacturing development of energy storage systems dedicated to the transportation sector.
Nearly all money was used in equipment and process development of manufacturing systems where the product was not well defined and in some situations where a "product requirements document" did not exist.
DOE excluded many good manufacturing approaches from the $2B grant money and at the time companies did not see significant adoption of energy storage in the automotive industry. Most of the automotive companies did not have a concise definition of their requirements, and some did not even believe in electrification of the transportation sector.
Grant recipients did not recognize the lack of energy storage market for automobile applications or the potential in the stationary sector. It is easy to be a "Monday morning quarterback coach"… The US lithium-ion manufacturing failed because of unnecessary marketing races, political maneuvering, and ambition forces.
The DOE ability to develop a robust "risk analysis" program for dissemination and evaluation of the grant application assumptions and projections was not stellar. OK, maybe it was done in the name of "recovery" and money is not so hard to print.
However, the ability of the grantees, to plan and execute the awarded grants, did not follow a scientific, data-driven approach. The product development effort did not take into account the application requirements and the entire product lifecycle.
I remember the funny arguments in the definition of what constitutes large format cell when in reference to the development of financial means towards the disposal of lithium batteries at the end of their useful life. Still not sure if the arguments where about the lack of knowledge on what to do with the batteries or that the size of the cell actually affects the disposal effort.
Today, there is a small but tenacious adoption movement towards electrification of transportation, and we have adequate information on the importance of energy storage for the implementation of large-scale energy collection mainly solar and wind farming.
The need for sizable energy storage systems increased substantially. Around half million electric cars are in operation and DOE projects one million by the end of the current year. Probably we will not reach the projection numbers, but the reality is that there is substantial demand. We must also recognize the need for bulk energy storage systems and the interest in lithium-ion to satisfy the need.
On the surface, the price per KWh of these systems also decreased substantially. However, we are aiming at cutting the current cost in half in order to compete with other energy storage solutions or provide an economic viable product.
In the quest to the goal of reducing the cost, we should analyze what contributes to the gains from early 1990's initial product introduction to today's state of the art lithium-ion products.
· The commoditized 18650 lithium cobalt cell experienced more than a 3X improvement in energy capacity due to material improvements and manufacturing optimization. Manufacturing yield improvements, better tooling, better quality / energy density materials, and others significantly impacted the cost.
· The price of raw materials (i.e. Cobalt) has been relatively stable. There are plans for substantial improvements in the throughput capabilities of lithium, cobalt, and manganese materials production. However, the majority of these efforts are located in countries with significant political and security volatility and in the absence of competitive alternatives, significant sustainable cost reductions are doubtful.
· Government funding programs provided substantial aid to manufacturers. The programs offset the capital equipment and R&D costs. If these incentives are eliminated, the cost for CAPEX development has to be recovered somehow.
· Lithium-ion battery cell manufacturers do not make substantial profits. The industry is very poorly integrated where the raw materials suppliers and the systems integrators realize anywhere from 5 to 35% profit, but many of the cell manufacturers lose money.
The most desirable characteristic of a battery is cycle life. Lithium-ion has a much greater cycle life than deep cycle lead acid batteries, yet the legacy products that wholesalers offer at around $175/KWh still dominate most renewable energy applications.
Legacy products have at least 3X lower cycle life at depths of discharge greater than 40%, are sensitive to temperature, require substantial more maintenance, have poor efficiency, and offer reduced control.
So why is a legacy product preferred for applications requiring 80% plus depth of discharge at one-hour rates where efficiency is critical?
Just like in the motive applications, the issue is that the system design/configuration has not been designed to project future needs of large energy storage systems. Potential customers for large energy storage systems, in general, do not view the current solutions as mature enough to risk full commitment. Most of the technology adoption to date in the bulk energy storage sector has been mandated or motivated by political expedient.
With regards to the current lithium-ion production process, we could list the parameters that influence the life cycle of a cell and have some impact on the system's safety. The parameters below, when not controlled within a narrow distribution, create difficulties to assemble a robust system. The difficulties increase if the system requires arrangements with parallel-connected cells that can't be independently monitored and isolated.
Critical parameters in the manufacturing of lithium-ion energy storage systems:
· Variation of internal resistance.
· Variation of porosity and structure of the separator.
· Management of corrosion inside the cells.
· Robustness of the terminations ampacity.
Internal resistance variance within a lot or between lots is often dependent on the initial steps in the manufacturing process, mainly incoming inspection of raw materials, mixing procedures and coating parameters. Regardless of the degree of process characterization and definition of process dependencies, the cell performance is measured in defects per unit area. More, for a particular process Cpk the variance always increases with the increase of area in a cell.
If we are to believe that the difference in the cell's internal resistance is inversely proportional to the system life cycle, and large area cells or electrode pairs tend to have greater internal resistance differences than there is a limit to the area of the cell or electrode pair.
The same concept may apply to "specific area capacity" (thickness of the electrodes, specific energy of the materials, and other capacity related parameters).
"Poly based" separator materials for the most part shrink when exposed to heat. Some shrink more than others, but at 90 °C, some separator materials have a 3% or more shrinkage. The drying process along with the tension of the material in the assembly can create significant variations in the porosity of the material leading to variation of current densities. The porosity variation, in my opinion, can exuberate the temperature variation on the coated surfaces. In extreme cases, the porosity variation can be such that provides preference for plating growth.
The larger the area of the separator the more process control is needed. Therefore, cells with small electrode areas are preferable to cells with larger electrode areas. There are methods of validating the integrity of the separator before filling the cell with electrolyte, and the cost and difficulty level varies with the type and size of the cells.
The conditions of the cell assembly at the electrolyte introduction and formation are of extreme importance to limit formation of corrosive agents that degrade the cell over time. There is an array of methods to minimize or remove gasses after the cell formation process.
The larger the electrode volume, the more difficult it is to condition the cell and the greater the probability to generate corrosive agents, therefore, the preference for building cells with a lower electrode volume.
With today's technology and data management, 100% cell characterization is possible. One can separate cells in different characteristic's bins and assemble modules and systems according to some cell control philosophy.
From the perspective of product performance, the logic above suggests that the most robust lithium-ion energy storage systems comprises a string of single cells in series. Those cells should have an optimum variation of internal resistances, integral separation of electrodes, absent of corrosive agents with robust electric bus. The overall performance of the system depends on the performance of the worst performing cell. In this situation, one can attain full monitoring and optimum cell balance of the system. If one needs a 340 Vdc, one needs around 90 cells in a series string at 3.7 Vdc per cell and 107 cells at 3.2 Vdc per cell. The energy capacity of the cell dictates the final energy of the system. A 3.1 Ah cell would provide for a system with 1 KWh and a 100 Ah cell would make a 34 KWh system.
Since the level of difficulty to make single electrode pair cells in the 100 Ah capacity is probably beyond the economic feasibility for any application, one needs to parallel cells into cell blocks. When one connects cells in parallel, one, no longer can economically monitor and balance each cell. Here one needs to manage cell variance during manufacturing to optimize performance and mitigate safety risks.
A publication in the Journal of Power Sources by Radu Gogoana, Matthew B. Pinson, Martin Z. Bazant and Sanjay E. Sarma, titled "Internal resistance matching for parallel-connected lithium-ion cells and impacts on battery pack cycle life" demonstrates some of the issues that may arise if blocks of parallel-connected cells are not build within a narrow distribution of internal resistance.
Cell internal resistance is only one of the parameters that cause issues with parallel-connected cells. The termination and the ampacity of the cell connections to the electrical bus are problematic and of a large distribution. Just like the resistance, the difference in terminations ampacity can cause 10 °C or more temperature variation between cells.
Looking at the future of the development and particularly at development efforts that reduce the cost of lithium-ion based energy storage systems one must consider:
1. The improvement of system's cycle life.
a. Ability to produce cells with very narrow distributions of variance for critical performance parameters. One benefits producing cells or electrode pairs of relatively small area to minimize the cell-to-cell differences in current densities, electrode resistance and other parameters that are area related (defects per unit area).
b. Ability to provide robust cell thermal management.
c. Ability to monitor and control the cells.
2. The cost of manufacturing of the overall system.
a. Reduce part count. Reducing or integrating components provide simplification, increase reliability and improve throughput. Choosing development paths, that contemplate reducing component count, yields low-cost manufacturing in the long run. We should not limit the development to the elimination of caps, cases, and seals but to the entire electrode configuration where one has an ultimate goal of producing solid state electrodes. The drive to such goal requires integration of the discovery with continuous improvement in the manufacturing development.
b. Optimize yields. Currently, there is an estimated 30-40% yield loss in the materials costs and estimated 25-30% yield loss in manufacturing cost suggesting that the low hanging fruit for cost reduction is in the yields improvement.
c. Elimination of VOCs. The use of NMP as the carrier solvent of choice for slurry production presents serious challenges for the distributed cell manufacturing as well as for centralized, vertical integrated manufacturing. For the distributed production, one has to consider the Capex and Opex to comply with local ordinances, emission and personnel exposure limits. In mass production operations on needs to consider the massive investment required to supply the operations. I estimate that the giga factory requires 22,000 liters of NMP per shift and have to deal with 4,000 liters of NMP waste per shift with a production rate of 35 GWh/year.
d. Deployment of automation. In some areas, the cell assembly process is adequately automated while others lack operation-specific development. The lack of application-specific methods is the principal cause of the significant yield losses cited above.
e. Transportation. In-plant and intra-plants movement of the cells represents a significant cost in the manufacturing of energy storage systems. In a process where there are such high yield losses, one cannot be very efficient if transportation is not an integral components of the manufacturing process.
3. Safety throughout the system lifecycle.
a. Process validation by characterization of process dependencies.
b. Data-driven quality verification.
c. Statistically thinking approaches throughout the manufacturing processes.
4. The management of product disposal at the end of life.
Prismatic large format cells provide:
· Most flexibility in manufacturing to develop application-specific energy storage systems with tailored energy capacity and voltage levels. The best design in energy storage systems is the one that provides adequate power to meet the functionality of the application and provides the most energy. Cylindrical cells offer limited flexibility to provide the best energy density, temperature management, vibration, safety and tooling cost.
· Best methodology and efficiency for product validation prior to electrolyte filling.
· Best methodology and effectiveness for removal of corrosive gasses after formation.
· Least part count for a system able to operate at ambient pressure.
· Least complicated methodology to create robust ampacity in the electrical bus.
· Most potential for efficient thermal management.
· We also need to consider systems management and control, both during operation and charging. An adaptive control system is always better that active or passive control systems. Adaptive control systems are not easy to implement with particular pack configurations.
· The product development path to accomplish most of the above
Many claim that, in time, major reduction of cell production costs renders battery price a non-issue. I fail to see where such reductions can occur if we continue to use the current manufacturing methodology and materials.
As I stated many times, investing in lithium-ion manufacturing is risky because the current methodology has nothing novel to offer and most companies with extensive converting expertise can dominate the market overnight.
I am biased towards large format prismatic cells and believe in lithium iron phosphate chemistry.